Dual Rad Love

[BrianW]

Well I got my Rads from this Junk Yard in New York. They cost me $28.00 each. The shippimg was free and it only took a week to receie them.

They went from this:



To this:



Then finally to this:



If you wanna see more images: http://thengc.net/images/watercooled...ics/Radiators/

I still have to finish sanding, and then paint them.....

Brian W

[gmat]

Nice.
BTW how did you make sure that water goes through *all* channels and not just the middle ones ? (ya know, using only the 2 center tanks on each side)

[BrianW]

These are single pass radiators. They go in one tank straight to the other tank. See the original unmodified radiators.... What I have done with my modifications is make one large single pass radiator w/ the following fim dimensions: 9" X 12" X 2".

Brian W

[Rayman2k2]

what case you got?...and how many wb's are you running?

[TerraMex]

I dont think that's what gmat meant. I think its this : (love that photoshop).



Unless you're mouting them horizontaly, and use gravity+pump to use all channels.

[gmat]

Yep thats more or less what i meant. Water tends to take the shortest path - i don't know how these rads are made but i suspect you'll have actual flow in only 50% of the channels, the ones at center.
I may be wrong though i don't know how channels are connected inside.

[BrianW]

They will be in this case(still under construction):



I will be running 3-5 waterblocks....

This is how the radiator will sit in the case:

[gmat]

(yep i've seen your project on these forums)
why not using two inlets and one outlet, or the opposite (2 outlets, 1 inlet) so you force the flow across all channels ?

[BrianW]

Even if I use two inlets or two outlets the flow would be the same, I would just be splitting it at a different point. Either way it is after and before the radiators...

Brian W

[TerraMex]

I see. Nevermind then, i think you wont have problems with that configuration.

Actually, i have an old Olivetti Desktop case that's perfect for that design ... my old trusty (and huge) 386.


It screams "modd me" all over it.

[BrianW]

Ya I am trying to bring the ol' Desktop cases back into popular domain. Actually I want to be hte only person @ the next lan w/ a Desktop Case..... Hahaha

Brian W

[gmat]

Originally posted by BrianW
Even if I use two inlets or two outlets the flow would be the same, I would just be splitting it at a different point. Either way it is after and before the radiators...

Brian W

Hmm you didn't get my point.
I mean, using two outlets on each side of the output tanks, instead of a central one. That way the water flow would cross a larger area through each rad. Currently i suspect that central channels will get 90% of the flow.

[BrianW]

The resistance through each tube is the same, so I would imagine that each tube would get very similiar flow. Why would 90% of the coolant go through half of the channels? The coolant wants to go the path of least resistance, which means it wants to go through all of the tubes at the relatively same speed. The big open tank takes a volume of coolant, then the coolant needs to go somewhere. It will divide evenly between the tubes. That is how I see it.

Keep in mind that I did not change how this rad works in normal operation.

Thanks for the input, and keep it coming....

Brian W

[gmat]

When all other things are equal the path of least resistance is often the shortest. That's why i'm suspecting that this configuration is not using 100% of the resources of these rads. Water in outer channels will 'stagnate' relatively to center channels,which will see the highest water velocity.
That's the usual 'array' problem that Axle faced with his passive rad, BTW.

water in>--+-------+-> water out
           |       |
           +-------+   (1st parallel pass)
           |       |
           +-------+   (2nd parallel pass)

in this example the 1st pass will see most of the flow, while the 2nd parallel pass will see far less flow, because water has to take a longest path - thus a path of higher charge (=flow resistance).

water in>--+-------+
           |       |
           +-------+   (1st parallel pass)
           |       |
           +-------+-> water out

In this one each path has the same length, and will see the same flow.

[BrianW]

I undestand the reasoning behind that belief. I am under the assumption that the large pressure difference between the tanks and the tubes will maintain relatively similiar flows between the individual tubes, however I also believe the tubes on the outside will have slightly slower speeds. I do not think this will matter much as the water flowing slower through the outside wil have a littlw while longer to drop it's heat off. On this last pont I may be way off base.

Brian W

PS: I wish someone would of brought this discussion up earlier when I was planning the build....

[BrianW]

Finished sanding it!




If you want to see more pictures...http://www.thengc.net/images/waterco...exi-neg-press/

BrianW

[Obitus]

That looks great. I need to bite the bullet and buy a heater core that is big enough for 2 120mm fans, but I am just not sure what to get, any suggestions?

[g.l.amour]

that is a great job brian. u r going slow but steady. did u use some sort of copper polish to get the gunk off?

[BrianW]

Thanks g.l.amour. No polish, just some wet/dry sand paper. Thats only 150 grit, tomorrow I'll finish with some higher grit.

Brian W

[g.l.amour]

well, in my opinion you will be very happy when u got that system up and running. it is always considered overkill, but lets see, sometimes i forget to give my fans a voltage boost at startup. so then at passive cooling, my H2O temp goes from 21 -> 35°C . that doesn't even evoke my bios emergency warning (that should read diode at 50°C). so, ok, u do'nt need 2HC'z, but u got a bulletproof cooling system in that case.

edit: and this is tested at 15*167 @ 2.075V

[gmat]

Originally posted by BrianW
On this last pont I may be way off base.

Yep it's a common misconception. In substance, trying to cool down already cold water is not very effective. Or if you take it the other way, what is good for heat transfer in a waterblock is good for heat transfer in a rad: lots of surface area and good velocity.

Originally posted by BrianW
I wish someone would of brought this discussion up earlier when I was planning the build....

i'm not Batman i cannot be everywhere all the time ("holy thermometer Batman ! this guy won't use his heatercores at their max efficiency ! - To the batcave Robin, quick !")

Anyway, nice setup, you're going to make some waves at the next LAN with your 'desktop'.

[Blackeagle]

Wow Brian,

Well, now I know why you picked that rad. When you showed one of those on the thread about single pass rads being better I wondered what your set up was going to look like.

And it looks really strong. With two rads of that size you can hook up what ever blocks & TEC's you want.

What pump will you use to dive dual rads with 3-5 blocks?

Edit:
My mistake, it was Nitrix that posted one that looked much like yours on that thread.


Gmat,

I have a 2-342, about the same size as one of Brian's but with inlet/outlets at diagonal corners of the full faced side. How much do you think the differant locations of the inlet/outlet will help full flow? At least I think that is what you were getting at when suggesting dual inlets and single outlet.

And would not the pump choice used in Brian's set up have a effect on the flow in the tubes? A example being a 1046 being a low flow/pressure pump would this make it easy for the majority of the water to use only 3-5 tubes? Would a higher pressure pump (MD-15 or even up to 30Z?) tend to minimize the problem by driving the water to flow through all tubes to find a overall balance of pressure? I know it would still happen to a degree, but would not higher pressure reduce it (%) as velocity increased in the tubing? And how large a pump would be needed for such a large rad area combined with 3-5 blocks?

U-Hual for lans????

And why do I feel this growing temptation to hack saw a tank on a single pass rad????

[gmat]

Originally posted by Blackeagle
How much do you think the differant locations of the inlet/outlet will help full flow?

How much, honestly i have no idea. I'm just confident that it *will* help as the effect can only be positive.

Originally posted by Blackeagle
And would not the pump choice used in Brian's set up have a effect on the flow in the tubes?

Of course yes. But the distribution of flow (i.e. the overall picture) won't change, center tubes will get most of it, though the minimal flow will be higher. It depends a lot on channel geometry, their size vs inlet / outlet sizes etc.
There won't be any balance even with an Iwaki. And in any case he won't be using 100% of the rads capacity. With an iwaki though, things will be pretty close to 100% as most rads (according to Bill's data) do not benefit much from uber high flows.

[Blackeagle]

I guess I'm missing it Gmat, regarding the percentage of flow per tube. And I could have stated my thinking better in the first post.

My thinking is based on the way pressure/resistence goes up as a square to increasing velocity in tubes.

Thus a 1046 with it's low flow and low head would allow the flow to all go through a small number of tubes. With a system having dual rads and 3-5 blocks flow rate would be well below 1gpm. So here I agree the flow goes through a minimum number of tubes. The velocity per tube would be to low to overcome the resistence differance caused by the added distance to travel in the two tanks.

A MD-20 Z with 7X the head can increase flow in his system to around 2gpm. The incrased velocity would create 5 to 6X the resistence, which would override the slight resistence differance between tubes resulting in a more balanced flow as the resistences caused by velocity in all the tubes would be greater than the differance in resistences caused by increased distance traveled in the two tanks.

Would still not be completely balanced as the distance in tank travel is still there. But that travel factor is now a lesser secondary resistence compared to the resistence from the velocity in the individual tubes.

What is my error or oversight?

My temptation is growing. What would you think about cutting away one brass tank, then replace it with one of plexi? With a outlet at each end with caps for each. Could then test to see the differance caused by diagonal inlet/outlet by alternating which is used. Could test pumps as well. But how to messure differances in flow per tube? Best I can invision is to add a dye to the water and then observe the flow of the dye through that plexi tank. But that is still only a subjective visual observation test. While repeatable, no way to get accurate messurements.

Ideas?

[gmat]

Originally posted by Blackeagle

Would still not be completely balanced as the distance in tank travel is still there. But that travel factor is now a lesser secondary resistence compared to the resistence from the velocity in the individual tubes.

We're just saying the same thing, different ways
Not balanced still, and i maintain a 'diagonal' setup would be better overall. And not everyone can afford an Iwaki...

Originally posted by Blackeagle

My temptation is growing. What would you think about cutting away one brass tank, then replace it with one of plexi? With a outlet at each end with caps for each. Could then test to see the differance caused by diagonal inlet/outlet by alternating which is used. Could test pumps as well. But how to messure differances in flow per tube? Best I can invision is to add a dye to the water and then observe the flow of the dye through that plexi tank. But that is still only a subjective visual observation test. While repeatable, no way to get accurate messurements.

Ideas?

This kind of test has been done for waterblocks, try a search on CFD on this forum. Same setup (parallel channels and central barbs) - you see the flow is not uniform across channels.
Apart from CFD, yes another way is to put a drop of ink and follow the "cloud" it forms.