Head Limitations in Closed Loop Water Cooler

joesgarage11 · Post by **joesgarage11** » Sun Apr 27, 2003 12:51 pm

I'm starting work on my second water cooled computer and had a question for the group.

In a closed loop cooling system with no resevoir except the volume of water in the system, does the pump suffer the same reduction of flow caused by the effect of head?

My instinct says it does not reduce the flow. I'm thinking the force of gravity on the water in the system on the way up cancels the force of gravity on the water on the way down. Also, the force of the pump drawing water in should equal the force of pushing water out. Since a closed system wouldn't be affected by air pressure, that should not be a factor. It seems that the only way to check this would be to check the flow
with some kind of flow gauge while varying the head height. Any thoughts?

The reason for this speculation is that I would like to run a linear radiator vertically to take advantage of the chimney effect. It fits better into the room that way, too. See below for radiator details.

My first water cooled computer gave good, but not great results with a homemade copper waterblock, Danner magdrive pump, and a 120mm fan mounted to a Toyota Supra heater core. I pulled it after running for about a year and went back to air cooling. The noise is getting to me, so this time around I'm going for quiet. I realized that the heater core was restricting the airflow too much for the fan to push much air through it. For better efficiency, it needs a less restrictive radiator.

The plan:

Home built copper/acrylic waterblock on 1.4 Ghz Athlon
Home built copper/acrylic waterblock on Geforce 2 GTS/Pro video card
Home built copper shroud on 80Gb Barracuda IV hd
Home built radial fin radiator
Water/air cooled power supply
Danner magdrive pump driving it all
120mm fan blowing on the radiator

By now, some of you may be thinking "Dood, why don't you just buy the parts?".

I just enjoy the designing and like to spend as much quality time with my power tools as possible.

Here's a look at one of the designs for part of the radiator. It's made with a 28" length of 3/4" copper tube with eight 24" long copper fins 1-3/16" wide. Next to it is a detail of the fins on the first prototype. The fins are held on with zip ties. The finished radiator will probably have four of these units in a box.

That's all for now. More posts will come as the project proceeds. All comments are welcomed.

Bat · Post by **Bat** » Sun Apr 27, 2003 5:40 pm

You're right about the head: it's not an issue.

Jan Kivar · Post by **Jan Kivar** » Mon Apr 28, 2003 4:54 am

That would be a concern if the pump "sucks/draws in" water.

Cheers,

Jan

jinu117 · Post by **jinu117** » Mon Apr 28, 2003 11:10 am

I can't believe... what these peeps saying. Of course the head will be an issue just like any other restriction in system. If the world is ideal and energy is conserved, sure why not? Why do people mention about difference between spiral block and white water about restriction? Head is basically how much restriction can the pump overcome... In your case, it probably doesn't matter much but something to think about.

pingu666 · Post by **pingu666** » Mon Apr 28, 2003 1:50 pm

its a good question and i dont know
itd have less effect i think, cos u are force feeding the pump

joesgarage11 · Post by **joesgarage11** » Mon Apr 28, 2003 3:51 pm

Thanks for your replies.

Jini, I believe head in an open system such as a pond pump is a combination of air pressure acting against the water in the tubing and gravity pulling the water down. In a closed system with no air pressure to deal with and an equal volume of water going in to and out of a radiator loop, don't we achieve a balance? There will still be flow losses from the length of the pipes, multiple pipe fittings, waterblocks, etc.

The project will continue. When it is done, I'll have to check the CPU temps with the radiator in different positions to see if it has any effect.

I made more fins for the radiator today. Need 64 in all. Next topic is a description, concept drawings, and photos of the completed radiator unit. It needs a name, though. Picture a vertical box 8" on a side and 48" high. The Chillin' Chimney? Tower of Cooling Power? The Chilla from Manila?

crisspy · Post by **crisspy** » Mon Apr 28, 2003 6:24 pm

I feel that I can say with authority that head is not an issue here. In fact what you need is to have an open resevoir at the top of the loop, so that you don't have any risk of an air lock in the top of the system (at which point head would be an issue). That is how everyone does woodstove heated water: big tank up high, pump (or convection forced) down low by the heater. In your case, the convection forces in the water will be negligible, and you should pump the hot water up to the top, and let it cool down through your radiator on its way back to the computer. That way you get an efficient cross-flow heat exchanger between the air and the water. The water coming out should be just slightly warmer than the air going in. If you try to do it the other way around, you can only get to the halfway point in temps between the air and the water. Google 'cross flow heat exchangers' if you want more info. If there's any issue with pressure drop in the lines, then make them bigger.

Nice job on the radiators by the way. They look really good. How did you thermally bond the fins? If you didn't, I would suggest that you can get thermal paste for refridgeration applications that would work well, and be fairly cheap, although solder would obviously be the best.

Cheers, Crisspy

joesgarage11 · Post by **joesgarage11** » Tue Apr 29, 2003 10:47 am

I haven't quite decided how to bond the fins. It's a lot of area to cover. The fins are radiused where they meet the 3/4" tubing and I'm leaning toward keeping the zip ties. Maybe soft solder, but I've got to get a smaller torch before trying that.

Bat · Post by **Bat** » Tue Apr 29, 2003 1:51 pm

joesgarage11: As an alternative to using round tubes with fins on them, you could get a similar amount of copper area by using more tubes, each one squashed nearly flat. I suppose it might be best to have them running across the box, on their edges. If you do it that way, then you'd get best results by having as many tubes as you can afford (in time, weight or money) near the bottom of the chimney, and making the chimney go right up to near the ceiling.

crisspy: The direction of water flow through the heat exchanger will not matter. What you say about counterflow vs. parallel flow is true of course, but only if the fluids flow through the exchanger so slowly that they end up near equilibrium. In this case, the water temperature will change by less than a degree: you can work it out from the flow rate (in kg/s), the CPU's power and the specific heat capacity of water. The water temperature will be close to constant everywhere in the system, and several degrees hotter than the room air.

crisspy · Post by **crisspy** » Tue Apr 29, 2003 1:51 pm

Hey joesgarage (like your nic btw)

If you replaced those zipties with copper or steel wire, you could then solder the fins right where they sit. I've done lots of big stuff soldering, and it's not hard to do. It shouldn't be much harder than house plumbing. You would want to use standard plumbing acid flux paste, coated onto the fresh sanded pipe and fin-bottoms, wire it together in place, heat it up, and start feeding in the solder when the flux bubbles out. The solder should fill any gaps by surface tension effect, and if you don't add too much it won't even run out very much onto the fins. I would also suggest using a tin-lead (60/40?) solder because it will melt at lower temps than the normal modern lead-free stuff used for house plumbing. Also, sometimes when you are doing large area work like that, you need to freshen up the flux as you move along, so I would have a small brush handy to put on more where needed.

To Bat: I didn't do the math. Thanks. Your figuring sounds about right now that I think about it. Nice that water cooling handles these small heat loads in such a non-critical fashion.

joesgarage11 · Post by **joesgarage11** » Wed Apr 30, 2003 8:03 am

I'm going to take crisspy's advice and solder the fins on. The copper elbows for the radiator with four tubes are already soldered on with lead-free plumbing solder. I'll use soft solder on the fins which will melt at a lower temperature and should not loosen the elbows.

The zip ties are a lot easier to work with, but the solder should give a much better heat transfer.

I did more concept drawings of the finished tower last night. Should be done with them tonight. The plan is to do the fins this weekend if the weather is good. The soldering is an outdoor project. Tried soldering in my indoor workshop, but it left me a little light headed and a strange taste in my mouth.

jafb2000 · Post by **jafb2000** » Thu May 01, 2003 3:52 pm

Solder the fins or use a Mocal/Earl motorcycle oil radiator/cooler, or
the sandwich plate ATX coolers you can buy from NAPA etc if in USA.

Zip ties will not give sufficient contact re heat transfer area at the
interface even remotely. You have no head in that system as long
as the system is fully primed with liquid before use AND you ideally
use a reservoir on the top to allow air to be removed from in-circuit.
Basically like a swirl pot on some racing engines to remove air and
cavitated air from circulation rather than recirculating it.

If water cooling a PSU, invert it so any leaks drop away from 300V.DC.
A PSU (SMPS) often has the primary-side mosfets electrically connected
to heatsinks, which will be 300V.DC - often but not always but assume.

Frankly, a half decent BIG extrusion for an ATX PSU could easily make
them fanless. Fit a 1U 300-350W PSU in base (re-PCB-routed for shape),
so allowing a very big 140x140x46mm high-finned extruded heatsink.
Thus water cooling should not be required for a PSU, plenty of big SMPS
in industry that are not air cooled - run hot, but do so truly silently.

The PSU fan was originally designed for case-cooling, and co-locating
the field-replaceable-unit into the PSU since their life are inter-related.

joesgarage11 · Post by **joesgarage11** » Fri May 02, 2003 12:57 pm

Thanks for the advice, jafb2000.

The resevoir on top of the radiator is a must have, but I haven't decided what to use and where to put it yet.

I've seen a power supply in another forum done with the large heatsink mounted out the back of the case. It looked great, but as I remember some of the other ps components heated up too much and the builder ended up sticking smaller heatsinks on them with hot glue

I'm planning on using a temperature controlled fan in the ps to ventilate the case and water to cool the ps regulators.

jafb2000 · Post by **jafb2000** » Fri May 02, 2003 1:04 pm

Re reservoir, non-metal is best re condensation.
Whilst acrylic may look tarty, it's not a bad application for it.

Same metal through-out removes the corrosion issue, but
mixing (say) steel with copper, or alloy with copper = depletion
of coolant anti-corrosion additives within about 9 months or so.

Same goes for cars, and nearly all have dissimilar metals,
even alloy heater-core & block/heads /very/ often have a
steel pipe or similar somewhere in the coolant system.