Ideal clearance for fans on radiator?

I am going to be pulling 3 Sanyo Denki 1011's through a PA120.3 radiator (thickness approximately 2.5 in). I am going to mount the radiator on top of my case, and realize I obviously need to allow for a gap to allow the fans to draw air through the radiator.

My question is this: ideally, how many inches should I allow between the case and the radiator to allow the fans to perform optimally? I was not able to find the proper phrasing for an adequate search, and would thus appreciate your input. Also, science > guesses, so I would love any relevant links that I could study.

I'd say about 1 inch to 1.5 inches. Give a "thumbs up", that's a pretty good distance. (length of your thumb)

Technically, you don't really need any shroud between the fans and rad, the radiator has a built in... .5" shroud? All of Thermochills radiators do.

Clearance for the case and radiator () is different. I left around 2 inches when I top-mounted my heatercore a while back. 1-2 inches should be good.

SolidxSnake said:

Technically, you don't really need any shroud between the fans and rad, the radiator has a built in... .5" shroud? All of Thermochills radiators do.

Click to expand...

Agreed. Thermochill cites a 2% difference between shroud and no shroud (wasn't clear on 2% of what, though). But my question was more geared toward...

Clearance for the case and radiator ()... I left around 2 inches when I top-mounted my heatercore a while back. 1-2 inches should be good.

Click to expand...

Thanks for the input, which is in line with NiTrO bOiE's response. I guess I was hoping for data on CFM based on clearance (though I certainly appreciate your real world experience!).

Maybe I could rephrase my original question differently: Has anyone had problems with throughput by NOT leaving enough clearance? If so, how close were you?

That question has passed through my head more than once.

By logic, you would consider a 120mm fan has an active area of a bit less than 113cm(square), 3fans ~339cm(square); The area fans are located on a PA120.3 is like 304x126mm, hence a perimeter of like 86cm, thus 339cm(square)/86cm ~ 4cm clearance.

That might sound a bit absurd, but that's what my logic points me...Area of exhaust = Area of input --> No air pressure differential; For sure there's a loss through the PA, and also a bit by momentum (horizontal flow to vertical one) but this is no rocket science

Can anyone think of that math and please tell me how high on crack and/or illuminated am I please?

pejcao said:

That question has passed through my head more than once.

By logic, you would consider a 120mm fan has an active area of a bit less than 113cm(square), 3fans ~339cm(square); The area fans are located on a PA120.3 is like 304x126mm, hence a perimeter of like 86cm, thus 339cm(square)/86cm ~ 4cm clearance.

That might sound a bit absurd, but that's what my logic points me...Area of exhaust = Area of input --> No air pressure differential; For sure there's a loss through the PA, and also a bit by momentum (horizontal flow to vertical one) but this is no rocket science

Can anyone think of that math and please tell me how high on crack and/or illuminated am I please?

Click to expand...

Now that's what I am talking about!

I have two questions regarding your proposal. First, how did you come up with that perimeter value (86 cm)? Also, if I follow your reasoning correctly, a PA120.2 would have a different optimal height since the area/perimeter ratio would be different, let alone that a single fan radiator. It is not intuitively clear why that would be so to me.

Also, your model does not account for the flow rate of the fans. Would it make sense that a very high CFM fan would require a greater height that a very low CFM fan?

Incidentally, I often find my crack level is directly proportional to my illumination level.

ixtapalapaquetl said:

First, how did you come up with that perimeter value (86 cm)?

Click to expand...

here: http://www.thermochill.com/guides/PA120-3.pdf (I subtracted the bottom and upper parts)

ixtapalapaquetl said:

Also, if I follow your reasoning correctly, a PA120.2 would have a different optimal height since the area/perimeter ratio would be different, let alone that a single fan radiator. It is not intuitively clear why that would be so to me.

Click to expand...

Indeed, a 2x120mm would have a third of the area, but just ~24cm less in perimeter, so its clearance would be a bit less.

ixtapalapaquetl said:

Also, your model does not account for the flow rate of the fans. Would it make sense that a very high CFM fan would require a greater height that a very low CFM fan?

Click to expand...

Right! I don't consider CFMs, just areas. That reasoning comes from the fact that any (ANY) amount of air going through the fan itself has to come from somewhere else. If that air comes trough an area just like the fans, It'll be like a plain duct connected to the fan (add to that rad's resistance + momentum loss), so having no effect on adding resistance on fans pull.

If a fluid goes through some area to a lower one, fluid increases its speed and pressure; The opposite is also truth. That's where that model comes from. EDIT: please take on consideration that air CAN an DOES comprises, liquids don't AFAIK.

At the end: a whole meter of clearance is fine, but AREA/PERIMETER seems to be a much better minimum clearance...Avoid resistance over fan's specs.

ixtapalapaquetl said:

Incidentally, I often find my crack level is directly proportional to my illumination level.

Click to expand...

I do too!!!, but right now I'm on like 7 beers, 5 to go...The only thing good of being unemployed: Crack, beers...and some bit of illumination

Wait!!! does that means the more illuminated you are, the more crack you smoke? or did you meant the opposite?

Sheers!

The idea of perimeter area is not bad.
Now what about Bernoullis law... A fan creates a low pressure on the suction side and a high pressure on the blow side. The air is then moving considerably faster on the suction side, does that mean you need more or less space on the suction side? I think less, would be interesting to hear what others think.

Theory is one thing, in real i think the difference between 2 inches and half an inch is very very small.

@pejcao: still trying to wrap my head around this. I noticed that for your area calculations you used the 3 circles, while for your perimeter calculations you included the entire rectangular structure. Why is this? It seems that you should be using the 3 circumferences instead. If not, you would arrive at very different values if, for example, the plastic surrounding the fans was increased say tenfold (thus increasing the perimeter). I see no physical reason why this would be the case.

Actually, now that I look more closely, you only used two of the sides, which makes even less sense to me - that would be more like (sum of areas)/(sum of diameters), which according to my calculations would be a constant. Could you help me understand?

Please bear with me, for I was but a math major and have virtually no understanding of the physical world around me.

th3 said:

The idea of perimeter area is not bad.
Now what about Bernoullis law... A fan creates a low pressure on the suction side and a high pressure on the blow side. The air is then moving considerably faster on the suction side, does that mean you need more or less space on the suction side? I think less, would be interesting to hear what others think.

Theory is one thing, in real i think the difference between 2 inches and half an inch is very very small.

Click to expand...

Woa! your crack dealer is way better than mine!

There's also the fans blades shape; On the pulling side it's almost horizontally cutting the air, on the other, almost perpendicular to airflow. That's why (at least a reason) a fan always works better pulling than pushing.

One thing I noticed when I had my fist wc rig (junkyard rad sitting horizontal at bottom of the case) was that after changing my case rubber feet (~4mm) for wooden ones (20mm) mainly because of dust reasons, was that the humming vacuum cleaner sound it made was gone. So believe me, too little clearance is BAD.

Anyhow, would be great if someone just comes with a model for clearance calculations (both, pulling & pushing, not like those formulas I've seen somewhere in this very forum about shrouds depth INSANE -add-crack-here- ROCKET SCIENCE).
I'm designing a homemade Alu angles case: rad horz. at bottom, in front of PSU (also at bottom) both isolated from the mobo and units area. So I need a clearance from the floor to case (feet) and from fans at top of rad to the roof of the isolation chamber. This thread really interests me, I still believe on the "perimeter" madness, but I don't want to end with a tall-as-me computer case.

Cheers!

dammnnnn you guy lost me totally and I was trying to figure out how far from my top plate to mount the radiator

ixtapalapaquetl said:

@pejcao: still trying to wrap my head around this. I noticed that for your area calculations you used the 3 circles, while for your perimeter calculations you included the entire rectangular structure. Why is this? It seems that you should be using the 3 circumferences instead. If not, you would arrive at very different values if, for example, the plastic surrounding the fans was increased say tenfold (thus increasing the perimeter). I see no physical reason why this would be the case.

Actually, now that I look more closely, you only used two of the sides, which makes even less sense to me - that would be more like (sum of areas)/(sum of diameters), which according to my calculations would be a constant. Could you help me understand?

Please bear with me, for I was but a math major and have virtually no understanding of the physical world around me.

Click to expand...

LOL, I'm terribly sorry, is just that I have to think in 3 dimensions, translate my insanity from Spanish to English, do some quick calculations AND keep my spelling as right as possible at the same time (while having beers and hug my bong, of course)

First, Yes, 3 circles on the fans side, that's fans own inner area, if you change those hypothetical 120mm fans for 140mm ones, their area will change.
Second, I did use both sides, look at the numbers
Third: (sum of areas)/(sum of diameters) is no constant my friend cm(square)/cm = cm! (but I can't seem to find where I wrote that formula you are referring to)

The bottom area (under rad, where air is coming from) is calculated by 4 surfaces and THE clearance. Air comes from both sides of the rads active area (~30.4 CM each) and both bottom and top of same active area (~12.6 CM). Top area = 3 circles, lower area = Rads active perimeter BY clearance ( 2x30.4 + 2x12.6 = perimeter)

Cheers!

extraoddsnends said:

dammnnnn you guy lost me totally and I was trying to figure out how far from my top plate to mount the radiator

Click to expand...

Go nuts with some powertools and make some holes to feed the fans, lowers your case temps and looks sooo much better. Im very satisfied with my current setup, made 2 big ugly holes on the rear part of the top of the case, then 2 aluminum frame fans to cover the holes, then a 40mm thick rad and finally 2 regular fans on top to complete the sandwich. Now i just need to change the pull fans to some nice looking alu fans as a final touch, gonna be very snacksy looking stuff. The rearmost push fan is veeery close to the PSU but the air seem to be flowing nicely.

Ok, perimeter is the rectangle made by the 4 colored surfaces over your case top. Perimeter Length x Clearance == Fans area (A1+A2+A3)
There's an image, so I save 1000 wds!

I've been through those calculations myself and agree 100%*. For a 3x 120mm rad you need ~1-1/2" clearance, for a 2x 120mm rad you only need 1-1/8". Not that it's a big difference.


*Note: The PA 120.3 is at least 36cm long (12cm x3), not 30.4cm. Thermochill has the main rad (not counting end tanks) as 12.4x39.9cm, yielding a perimeter of 104.6cm. Taking out 12.1 cm for each side (= 24.2 total) for a double rad yields 80.4 cm. So:

fan area / perimeter = clearance(min)

339 (cm^2)/ 104.6 cm = 3.2 cm = 1.3"
226 (cm^2)/ 80.4 cm = 2.8 cm = 1.1"


But there is a loss on the triple rad that isn't being counted. The center fan will end up in a slight vacuum compared to the two outside fans because it's only pulling from the sides instead of the sides and one end. I'd tend to bump the clearance a hair to 1-1/2" for the triple to account for this and minimize CFM loss unless you're also using higher-pressure fans. I'm using low RPM, low pressure fans on my heater core so I've bumped the clearance out to 1-1/2" for a double and it's working great ...

If you have a space behind the fans then the fans need to have baffles to prevent air being sucked backwards through the neighboring fan. Even a single fan has a pretty good flow of air backward through the space between the fan blades and the housing wall. If you completely block a fan you can feel a strong wind blowing thru that space. A small piece of cardboard that fits between the fans and extends down to the rad will do the job.

Here's a suggestion.............

QuietIce said:

*Note: The PA 120.3 is at least 36cm long (12cm x3), not 30.4cm. Thermochill has the main rad (not counting end tanks) as 12.4x39.9cm, yielding a perimeter of 104.6cm. Taking out 12.1 cm for each side (= 24.2 total) for a double rad yields 80.4 cm.

Click to expand...

You're right, sorry for my lack of professionalism taking measures. Now I feel confident about the "area/perimeter=clearance" model since now I know I'm not alone in this.

QuietIce said:

But there is a loss on the triple rad that isn't being counted. The center fan will end up in a slight vacuum compared to the two outside fans because it's only pulling from the sides instead of the sides and one end. I'd tend to bump the clearance a hair to 1-1/2" for the triple to account for this and minimize CFM loss unless you're also using higher-pressure fans. I'm using low RPM, low pressure fans on my heater core so I've bumped the clearance out to 1-1/2" for a double and it's working great ...

Click to expand...

I have thought of that too, but considering air has to travel through that "mini shroud" space, the rad fins, and from the "free flow" AKA perimeter clearance area, my guess (simply/practical thinking) Is that air pressure will equalize (being the same) over those 8 areas around the perimeter; It'll be like adding negative pressure vectors. I'm not saying pressure will be "the" same around perimeter, but at least "safe" to think of It that way.

Now I had thought before a "worst case scenario" of that issue you noticed, imagine a rad (perfectly square) and that you place 9 fans on it; four fans at the corners will pull from 2 surfaces, four at the middle-borders from one, the one fat the center from none...none? so no air is pulled from that fan?...I guess is indeed a Worst case scenario since It gives me a headache just to imagine all those vectors! but at the end this example notes that you are right, but again this is a worst case scenario.

billb said:

Here's a suggestion.............

Click to expand...

Thanks 4 pointing out the "point" in discussion!

Does anyone knows how to get sketshup working nice on wine?

billb said:

Here's a suggestion.............

Click to expand...

Taking that route puts it at 1-3/4"+ min clearance.

If I were doing this with my low RPM, low pressure fans I'd probably go that way.

pejcao said:

Now I had thought before a "worst case scenario" of that issue you noticed, imagine a rad (perfectly square) and that you place 9 fans on it; four fans at the corners will pull from 2 surfaces, four at the middle-borders from one, the one fat the center from none...none? so no air is pulled from that fan?...I guess is indeed a Worst case scenario since It gives me a headache just to imagine all those vectors! but at the end this example notes that you are right, but again this is a worst case scenario.

Click to expand...

That's a whole different design issue. In that case I would recommend a monster fan for the center and maybe a small air duct for it's intake. A simple plate (any stiff material) the size of the rad or a hair bigger with a 12 cm hole in the center and raised off the deck 1/2-3/4"" should do it (with another space ~1-3/4" above that for the other fans). Not only would you balance the intake a little but you'd also create a "jet" of sorts on the downwind side, which would provide a bit more vacuum for the other fans ...