Radiator Sandwich / Stacked Radiators / "SLI-Radiators" Test results

[HESmelaugh]

That would be very interesting. Do you have the radiators, or are you thinking of getting a setup like that?

[bigslappy]

howsa about a single shoud in-between rads that is a simple box / full cover size & have a push /pull fans on the outside of the stacked rads ??

[HawaiianSupermn]

I love the sandwich action buddy. I was considering this instead of a single p.a. 120.3 rad up top. I was also thinking of doing this with single rads in small spaces but wasnt sure how the barb setup would turn out. Nice Job!

[bigslappy]

instead of 90 degree's i'm gonna mod up some copper pipe that will take most of the hard turns outta the tubing runs with the 90's
STAY tuned i'll post pix when i get the setup done .
I like this idea & am gonna run with it ..
........i got a dozen MCR320 0n the shelf ............

with hi speed fans &/OR push/pull hi speed fans would not the placement of the hot side vs the cold side out recieving the coldest air ...
have a minimal effect whether it be hot OR cold ?? meaning series looped ....???
.......................... any thoughts ANYONE ??

cause if this works u can have 2 mcr320 & some hispeed Yate Loons for less cost than some of the high priced rads not counting the fans price & most of us could have preformace at a reasonable cost
YEAH the size would be bigger but FTW on size/room............ Preformace way out takes the CAKE than JUST looking pretty ..................TEMPS Rule

[Webby]

Bigslappy, the are two ways of looking at the order of the rads well 3 if you count parallel! I guess you know what they are but I will outline them anyway and then get to what I believe to be best.

So Rad1 has the hottest water (inlet), this is then passed to Rad2 and then back to the loop.

1. Cold air over Rad1 first
2. Cold air over Rad2 first
3. Cold air over stacked parallel rads

We will now sidestep to theory for a second.

Imagine 2 tubes one inside the other, one filled with hot water the other with cold water (the following is dependent on flowrates so we will assume they are the same for both hot and cold) so you have water at 10°C (cold water) and water at 90°C (hot water) and you want to cool the hot water down, in co-current flow (option 1 above) your hottest water contacts the coolest water (largest driving force - good) assuming you have an infinite length of pipe the coolest you can get the water is 50°C and as the cold water gets hotter the driving force shrinks so it takes longer and longer to reach equilibrium. If you had counter-current flow (option 2) then along that same length of pipe you could theoretically cool the hot water to 10°C and heat the cold water to 90°C the driving force will remain constant over the entire length of the pipe however it will take longer to cool the the hot liquid to say 80°C as the driving force would be less.

Stepping back to real life now the question we have to ask is do we when running under option 1 heat the air coming out of the first radiator enough that it significantly impacts on cooling on the second this is a difficult one to answer as it will depend on a number of factors but most importantly on the airflow rate in relation to the liquid flow rate, assuming that liquid flow rate is fixed its all down to fan speed. If we have higher fan speed we decrease the heat transferred to each volume of air passing through the radiator this increases the driving force. So this means that if you use high speed fans then the order of radiators is probably of little importance as the air passing over each one will be of similar temperature (option 1 may provide small benefits due to the higher initial driving force), if however you use slower spinning fans then you would expect option 2 to be the better performer as the temperature gradient over both radiators would be more consistent.

Parallel rads option 3 is really of no use when stacking unless you want higher total liquid flow rates generated from the lower pressure drop associated with this flow regime.

So my advice would be option 2 cold air cold rad this should provide the best average performance over a wide range of fan speeds. But probably worth testing to see what actually happens!

As it happens I have 2 PA160's stacked together in a push pull config Fan->Shroud->Rad->Rad->Shroud->Fan and I really should get round to testing them with the air flow the other way (currently using option 2 with 38mm fans ~1200rpm) but I have more pressing matters to attend to atm though that should be finished in a couple of weeks and then I will give it a go.

[HESmelaugh]

Sounds very nice, bigslappy! Looking forward to seeing what you can come up with.

Seeing that there were two tests on german forums that showed almost no temperature improvement from a single radiator to stacked radiators, I assume that it makes a significant difference, which rad gets ventilated first and which second.
More testing will give us more of a clue.

[Webby]

Right I have been thinking about it and decided that radiator order probably makes very little difference, my theory is right but due to the nature of the materials we use ie cooling water with air it just doesn't really make that much of a difference.

If a rad is dissipating 100W and the water flowrate is 1.5gpm then the water in water out delta is ~0.25°C, 200W = ~0.5°C and so on.

If you had an airflow rate of 50cfm and that was absorbing 100W then the air in air out delta would be ~3.5°C.

If you make the assumption that heat dissipation is linearly linked to air in water out delta (which from Martin's tests seems to be reasonable for this exercise at least) then if the first rad can dissipate 100W and raise the air temp by 3.5°C (say air in 20°C and water in 30°C delta first rad 10°C) then the second Rad the delta is 6.25°C and so the second rad can remove 62.5W (oh the joys of picking your starting figures).

So in theory as long as you can provide sufficient airflow to the rads to overcome the additional pressure drop then you should see a decent increase in performance otherwise you saturate your air and it can no longer remove heat from the second rad. Guess it will all be a balancing game!

[Martinm210]

This may be helpful in figuring out radiator order. This is the XSPC RS360 which is similar to the magiccool rads I believe, I think they are both a bit narrower than the MCR series (120mm vs 130+).

Anyhow, I recorded both air in and air out:


Somewhat surprisingly I used fairly consistently only about 50-55% of the air's capacity regardless of fan used. The more dens radiators were getting upwards of 80% or so depending on the actual air flow.

This appears to make sense in term of the additional 30-40% you gained stacking them.

With that I can only assume with the double thickness rads like a TFC, your net gain from sandwiching would be much lower because you've already used up nearly 80% of the air's cooling capacity with the lower fan speeds. Here is the TFC 480:


It shows the percentage of air capacity used was in the 80% range for low speeds down into the high 60% range for higher speed fans. That would mean stacking them probably is only good for about 5-10% gain low speed and maybe 20% or so with high speed fans.

It would be interesting to see though. Who's going to stack two 480 GTX's with nine mega high speed deltas and see how much heat they can dissipate..

I like the two stage cool order myself, second rad getting the coldest air in, but I wouldn't be suprised if in the end either order turns out pretty much the same. The biggest difference in stacking of the rad gain will be in the rad itself and fan combo, higher speeds should gain more, but it really depends on the rad. I wouldn't expect the double thickness rads to see much gain in stacking though, not enough air capacity left.

I'm curious how these even thicker rads planned to come out will perform. I wouldn't expect 100mm worth of rad thickness to be worth more than 10% over 50mm of rad thickness. The biggest gains are always in the frontal area, that's why many have gone to slightly over sizing the width or fan spacing, typically one extra row or around 130mm wide.

If I was going to design a rad, I would make one extra wide, but slim thickness. Maybe 160mm wide or as wide as possible to fit in most locations, and build a shroud into the rad for the extra shroud gain. I would also make the plenum chambers double side ported so they could be easily sandwiched. Material costs would then be relatively low, but performance would be fairly high and the stacking ability would make them modular. Maybe some sort of SLI specific fittings to sandwich them an easy means to attach both rads together with 25 or 38mm fans.

It's threads like this that might get them thinking more along the lines of modular sandwich/SLI type radiator setups instead of going extra thick.

[Freaky Freezer]

Really great test. I'll be looking forward to the difference in temperatures between the sandwich and 2 separate radiators. My guess is the difference will be less than 1 degree. At which point I'll slap myself for not thinking of doing the sandwich myself instead of bleeding and sweating, trying to get 2 separate BIP's into an unfriendly, mid-tower acrylic case. Although future watercooling upgrades now seem less painful for me. Thanks again.

[HESmelaugh]

Very interesting stuff. I am just now testing a Black ICE GT 120 vs. a GT 140. I want to find out how much benefit you get from that extra radiator width. According to Martin's theory, I might see some nice gains!

[Serpentarius]

Another important question!!

Can stacking reduce the fan noise as well???? Yes/No

pls check as well .. thx

[HESmelaugh]

Ok, I'll try to check that. Though I don't have any instruments for measuring noise levels. I'll have to judge purely subjectively...

[inCore]

I reckon the testing should find it's way towards the sticky here in the Liquid Cooling section.

[bigslappy]

The Rad i'd like to see come out of the Rad Mfg'ers is a Rad that will use the newer 250mm fan as those fans move alotta air & are quite for the amount of air they move & with the Mountiion mods & the like Cube Case systems this would be a great option as u could use the side walls as the rad systems & cover with a high pass thru air filters killing the dust & hide the rad all in the same unit & quite to boot ! Power consumtion would be less as you would have less motors . Even a Tower case could have the 250mm rad/fan system in the side covers .. my Xclio case has twin 250mm fans on it's side cover & with the filter mod i did it to this day has never seen a bit of dust in it & the air blows directly on to the MoBO ..
with a simple shoud system & a rad/fan setup to fit a twin 250mm fan high flow rad would cool quite nicely

[HESmelaugh]

Parallel radiator fail?

What do you guys think, will this work? How can I make sure that both rads are getting equal amounts of flow?

And yes, I am reviving this thread because yes, I have finally started testing radiator sandwiches again.

[Taylor01]

I was thinking about doing this for my htpc with two dual80mm. I might give this a try here soon. thanks for all the work you put in here guys.

[Spankyfart]

Parallel radiator fail?

How can I make sure that both rads are getting equal amounts of flow?

I don't think you can unless you have the exact same routing to the radiators.
And even then it will take the path of least resistance.

Put the radiators facing eachother, and then hook them up with that koolance splitter (I'm assuming that's what it is)

And yes, more rad stacking tests

[HESmelaugh]

This splitter is too long (unless it's modular) and so is this one.

But now that you mention it, I remember what I posted about how an ideal Rad-connector would look like:


And then I suddenly realized that I might be able to make just that with Delrin Ts and two of these! Thanks for the suggestion!

I'm testing the setup in the above pic currently, but I'll order the necessary parts and then retest with the simpler connection.

[mattkosem]

Parallel radiator fail?

What do you guys think, will this work? How can I make sure that both rads are getting equal amounts of flow?

And yes, I am reviving this thread because yes, I have finally started testing radiator sandwiches again.

I wouldn't do that. You want the rad that is closer to fresh air to be last in line, since the other rad will be receiving heat dump from the first. The rads should be in series, not parallel.

--Matt

[HESmelaugh]

Thanks, Matt. However, I'm only doing this setup for testing purpouses. I'll test several different options, including serial setups. And of course, I will post the results here.

[Webby]

Basically If you want to ensure equal flow through each rad you need to make sure that the path to each is the same so currently as you have it setup the rad on the left will be getting slightly less flow than the one on the right, how much will depend on what the difference in pressure drop is over the connections from the splitter to the rad and also the pressure drop flow relationship of the rad.

Basically for parallel flow a rotary T (as you suggest with the Koolance sliding fittings) is the ideal solution.

If I may make a suggestion with the testing measure temperatures in and out of each rad for the water and air.

My 2x PA160's which are stacked (soon maybe to be unstacked if I can fabricate a suitable new shroud) under load I see maybe a 5°C temp change over the first rad in air temp and then only a 1°C over the second, now this is potentially because I am not loading the rads up to anything like their max (I have a water out air in delta of ~5°C) but still it shows how much difference the hot air makes on the second rad.

[Ch3ck5um]

Perhaps I should sandwich my 120.3 & 120.2 together

[matsjuw]

Once again nice testing HES, looking forward to the new tests.