Radiator sizing for heat absorbtion

[[XC] gomeler]

So.. I'm gathering the last pieces for a chilled liquid project using a radiator to absorb heat. Problem is I don't know how effective radiators are at absorbing/dissipating heat. I want to be able to absorb from a sealed chamber around 600w of heat. So.. for such a setup in a traditional watercooling setup would you use a 2x120 radiator or a 3x120 radiator? Say you are dissipating 600w. Would 2x120 be drastically undersized or would it sort of work?

[[XC] riptide]

Interesting! I'm going to quickly say that I think heat will travel both directions equally well! Its a two way road. I think a 2x120 does ~300W in disapating from ambient of ~21C. Again... it all depends on the temp differentials and what your willing to achieve. What temps is the coolant runing through the radiator?

[MpG]

Since the radiator's moving heat by conduction, the heat should travel both ways just fine. Beyond that, it's going to depend a lot on your exact setup. Firstly, what kind of airflow is going to be going through the radiator fins, and secondly, what kind of temperature differential will it be seeing? The faster the airflow (i.e. the stronger the fans), the less radiator you'll need to move that much heat.

[[XC] gomeler]

Well I'm familiar with condensers/evaps in the 120mmx120mm size and they'd do just fine running partially flooded with refrigerant but I have no clue as to how effective radiators and water are. Figure the fluid is at -40C and has a relatively infinite capacity of dissipation(huge reservoir and cycling chiller). How big of a delta do you guys see between water temps and coolant temps with various loads? If someone can point me in the right direction/explain things I'd appreciate it.

On reflection I think I'm asking too much with too little information from the wrong crowd, I'll try to grab a used 2x120 or 3x120 radiator and give it a go.

[[XC] gomeler]

Since the radiator's moving heat by conduction, the heat should travel both ways just fine. Beyond that, it's going to depend a lot on your exact setup. Firstly, what kind of airflow is going to be going through the radiator fins, and secondly, what kind of temperature differential will it be seeing? The faster the airflow (i.e. the stronger the fans), the less radiator you'll need to move that much heat.

Since it'll be in a completely isolated box, figure the beefiest 38mm fans I can cram in there, so 150cfm per fan x 3. Sandwiching the rad should in theory improve airflow, correct?

[skinnee]

Here is what triples can do for heat dissipation at a 10C Air to Water dT.



As for sandwiching...check the MCR Stacked #'s. That is Rad|Fan|Rad setup, and really does not work as a rad sandwich. Slap another row of fans on it and the numbers will improve. I'm going to be doing further testing of the MCR320 Stack...there is performance capability just sitting there.

[Hondacity]

awesome graph .....i wonder how the koolance 3x120 rad look like

[[XC] gomeler]

Here is what triples can do for heat dissipation at a 10C Air to Water dT.

As for sandwiching...check the MCR Stacked #'s. That is Rad|Fan|Rad setup, and really does not work as a rad sandwich. Slap another row of fans on it and the numbers will improve. I'm going to be doing further testing of the MCR320 Stack...there is performance capability just sitting there.

This helps so much. Judging by your numbers and averaging all the results, a 2x120 radiator(figured 2/3 performance) with 2 fans pushing 240 cfm will be able to dissipate 600w of heat with roughly a 10 degree delta.

By sandwiching I meant using 4 fans on 1 radiator, fans rad fans. Figure that'd increase cfm by 40-60%?

[DeathWalking]

Since it'll be in a completely isolated box, figure the beefiest 38mm fans I can cram in there, so 150cfm per fan x 3. Sandwiching the rad should in theory improve airflow, correct?

Delta FFB1224XHE. I think you'll need serious sound insulation on that box, but they'll do the trick. They're the best fans they make -- I'm not kidding. The FFB1248XHE will do the trick too, but I don't know where you could find a PSU that puts out 48 volts -- you'll still need a meanwell for the 24 volts the other one needs. 264.62 CFM, and 43.31 mmH20. Makes a Triebwerk cry. Put double up the fans (1 on each side of the rad) and you're talking about some serious airflow/cooling.

[skinnee]

I'm going to be running push/pull tests to be certain as to the performance increase, just have to do these last couple i7 blocks first.

your estimate seems close, should be fine. Did you say you had a MCR220?

[Vapor]

By sandwiching I meant using 4 fans on 1 radiator, fans rad fans. Figure that'd increase cfm by 40-60%?

Helps more with low-speed fans...because pressure (at all airflow levels) scales with the square of RPM, high-speed fans already have a PQ curve that's pretty tall and doubling the height (what a fan sandwich does) doesn't help *that* much.

It does help somewhat though It's effectiveness is almost always higher per unit of noise than increasing fan speed is

[[XC] gomeler]

I'm going to be running push/pull tests to be certain as to the performance increase, just have to do these last couple i7 blocks first.

your estimate seems close, should be fine. Did you say you had a MCR220?

Don't have a radiator right now. There are a few used 2x120mm rads for $30 or less but I could get a MCR-320 for $55 on petras. Decisions decisions

Helps more with low-speed fans...because pressure (at all airflow levels) scales with the square of RPM, high-speed fans already have a PQ curve that's pretty tall and doubling the height (what a fan sandwich does) doesn't help *that* much.

It does help somewhat though It's effectiveness is almost always higher per unit of noise than increasing fan speed is

That does make sense. I should have taken fluids at school, I'd know such things

[skinnee]

if you can pick up a few 2x120's for $20...do it.

I'm still figuring the math on TEC's and efficiency at certain delta's on the hot side, but the lower dT you have on the hot side the coldside can pull more wattage...I think, but something tells me I am still reading through Jimbo's information wrong.

[Martinm210]

Here is what triples can do for heat dissipation at a 10C Air to Water dT.



As for sandwiching...check the MCR Stacked #'s. That is Rad|Fan|Rad setup, and really does not work as a rad sandwich. Slap another row of fans on it and the numbers will improve. I'm going to be doing further testing of the MCR320 Stack...there is performance capability just sitting there.

Awesome chart there, it speaks volumes!!!

Doesn't look like stacking is all it's cracked up to be, thanks for doing the tests. Having them all done on the same test bed is awesome.

[skinnee]

thanks Martin!

I'm running way past due on the test reports, I have all the data just not the review portion. Figured it will be released in full soon enough and the data needs to be shared...post the darn graph Cam!

[[XC] gomeler]

Grabbed a Black Ice X Flow Stealth GTS 250 for $30 with 1/2 inch barbs. Hope this works well enough. Bought two high speed 38mm panaflos, 110cfm. Will use this to start, if that isn't enough I'll throw another rad at it.

[Mech0z]

Here is what triples can do for heat dissipation at a 10C Air to Water dT.



As for sandwiching...check the MCR Stacked #'s. That is Rad|Fan|Rad setup, and really does not work as a rad sandwich. Slap another row of fans on it and the numbers will improve. I'm going to be doing further testing of the MCR320 Stack...there is performance capability just sitting there.

Must say IAm abit supprised of how well the TFC does also at high rpm

[skinnee]

the slightly higher FPI on the TFC gives it the advantage over the TC and RX at higher RPM.

[Mech0z]

Guess them being 75mm thick has something to say as well, just cant decide which rad I should pick when going gentle typhoon 1850

New TC120.4 when they are released
XSPC RX480 due to it being 25£ cheaper than the Feser
TFC Feser 480

[liguhy]

Here is what triples can do for heat dissipation at a 10C Air to Water dT.

+1 for the awesome graph. Really tells you everything about the performance difference among different density of fins. Wonder what adding a 2nd Y axis for noise vs rpm would tell. The 1400 merging point for performance is good to know!

[Martinm210]

Yeah, that chart is awesome!

I've had the same exact fans running on two radiators side by side for many months now. One side is a 480GTX and the other is a TFC480, I can not at all tell a difference between noise on one side or the other when they are turned up the same volts and thats with 4 fans on each, so you'd think it would be obvious.

So at least on my system, fin density doesn't seem to make any noise difference. Sure you'd have to run the fans higher to get an equal CFM level, but this chart is comparing RPM which should make noise pretty much equal at least in the sense of noises that matter (motor/fan chop/vibration).

[liguhy]

So at least on my system, fin density doesn't seem to make any noise difference. Sure you'd have to run the fans higher to get an equal CFM level, but this chart is comparing RPM which should make noise pretty much equal at least in the sense of noises that matter (motor/fan chop/vibration).

I had the impression that low-density radiators (Thermochill, Feser, XSPC) were optimized for low-rpm fans at <1400rpm not requiring high static pressure. For high static-pressure fans > 1400rpm, highly-dense radiators (HW Labs Black Ice) had a significant performance increase, but at a significant sound increase as well. Correct me if I'm wrong, but those are the two groups the 4 major radiator brands fall into for fin density/static pressure fans.