Maximizing Airflow in a Computer: Finding the Optimal Number of Fans

[KingNothing]

First off, I'm sorry if there is a more appropriate section for this. There is no section exclusively for fluid physics, and I am not really trying to engineer a prodct, so I posted this here. SO if you have to move it, go easy on me .

Think of a compute as essentially a box with some fans which can be on any face of the box, and blow either in or out.

My goal is to figure out ways to maximize airflow and minimize the # of fans used.

I think it might be obvious that having the box covered with fans would be the most obvious maximum airflow. However, that's not feasible for the application. I was pondering the possibility of just having two fans that are rated for 50 CFM (cubic feet per minute) on one side and just a hole or holes on the opposite side. I figure that the fans would create a low pressure inside the case, and then more air would naturally flow into the case through the holes.

On the contrary, part of me thinks that it would be better to have an equal number of fans exhausting air as fans that are intaking air. I think this because I figure "If I have two fans exhausting air, and air flowing through holes, to get inside, the fans will only be able to create a certain low pressure inside, and thus there won't be as much overall airflow through the case compared to having another two intakes."

However, how great would the difference be? Should I be looking for a design with more exhaust fans, more intake fans, or an equal number? I want the highest "Overall Airflow -to- Number of fans" ratio.

 

[CaptainQuaser]

Exaust is important especially at the back near the CPU, actually most computer do it fairly smart right now, anyways, there is obviously a power supply fan, some blow down, others blow opposite the plug. For gaming a videocard fan is important, there are fans attached to video cards, but you often want a side fan to keep airflow moving across the mother board around the video card. The rest is bonus, a nice big front fan I have seen before, to help keep air flowing front to back keeping the crucial CPU heatsink with a fresh air supply.

 

[KingNothing]

I see you are a computer nut like me. Not to sound stuck up, but we both knew that stuff before this thread was created. Crucial parts will always have more attention paid to them, but airflow efficiency itself in a closed "box" is what I am hoping someone will explain in a manner unconcerned with the 'important parts'.

 

[quark]

Fan flow capacity is not an independant characteristic and it is always governed by the resistance offered to the flow. For a fixed system, flowrate decreases with increase in system resistance. If you design your fan for 50cfm flowrate at a resistance of, say, 0.25" wc then as the resistance goes high, the flowrate gets reduced. The idea of going for a supply fan and an exhaust fan is to just overcome the resistance so that you can always ensure the required flowrate. If you carefully select a fan then using it for exhaust and making openings on the other side will be enough. Don't provide the openings just infront of the fan. If the fan is placed on the right, keep the opening to the left on the opposite side.

 

[Danger]

Wouldn't bottom=intake and top=exhaust be most efficient due to convection?

 

[Brad_1234]

So youre redesigning a computer? or making a new design for computers to sell?

First identify that a problem (really) exists. Why are you undertaking this? There must be some condition or cause, like its too noisy, its too much this or too little that. You have a computer but there is something wrong, define what the problem is.

Now just for talking, suppose you wanted to really cool the inside of the computer and applied something that created very cold air. It would cause moisture in the air to condensate, form water drops and get into circuitry. Then you would need to de-humidify the air, but that causes more static potential buildup which can also damage circuitry.

 

[KingNothing]

I am asking about airflow efficiency, not about troubleshooting a computer. There doesn't need to be any problem to learn about physics (fluid physics included). Please stick to the topic:

However, how great would the difference be? Should I be looking for a design with more exhaust fans, more intake fans, or an equal number? I want the highest "Overall Airflow -to- Number of fans" ratio.

Danger has established that airflow should be upwards. Say the box has one exhaust fan on the top. Would it really be worth it to add an intake fan on the bottom vs. just holes for air to flow through?

 

[Danger]

I wouldn't expect that an intake fan would make much difference to the airflow, and would in fact add a bit of heat itself from the motor. Once your airflow is established, it should remain at the cfm of the fan. If you have a fan at each opening, the one with the lowest cfm would restrict the other one to match its own capacity. Making sure that your inlet holes are big enough and not obstructed should be sufficient.

 

[Artman]

You are adding heat via the CPU and other components, perhaps you could design a stack vent to take advantage of stack effect on a small scale.

Stack effect calculations:

Q = 60 * Cd * A * sqrt(2 * g * (Ht - Hb) * ((Ti - To) / Ti))

where

Q = flow rate in cfm,
Cd = 0.65 (for unobstructed openings),
A = opening area, square feet,
Ti = indoor temp (Rankine),
To = outdoor temp (Rankine),
g = gravity.

Stack Effect Calculator

 

[KingNothing]

I don't understand the 'stack effect'...The link you provided is titled "Convective Flow due to the Stack Effect"...but it seems like it IS just convection. Am I missing something?

 

[Artman]

It is the tendency of heated air to become bouyant and lift itself without external power (fans). Some computers already use this process to cool their CPUs. A boiler stack, gravity ventilator, or chimney will draw air out of a room without a fan assisting the process.

Your post asked for ideas to give you the most airflow with the fewest fans. This way will give you air flow without any fans.

 

[russ_watters]

I wouldn't worry about stack effect, it is small for the heights and temperatures we are talking about. KingNothing, your idea is basically correct. But if you find that you are not getting the flow you want out of that 50cfm fan, the reason is pressure, as quark was saying. The main reason for having two fans in series (ie, one intake, one exhaust) is that you get double the pressure.

Propeller fans are not capable of generating much pressure, so if there is a lot of resistance (and 50cfm is a lot for that small of a space, so there will be lots of resistance), look into a blower. Blowers create a much higher pressure than propeller fans.

http://www.cooltron.com/technical_ChooseFan.shtml is a site that has info on computer case fan performance. Compare some fan curves for different types of fans.

 

[KingNothing]

This whole 'stack effect' has me confused. It appears to just be a synonym for convection.

How is a blower different from a propellor fan? Also, are blowers noisy?

Does 'double the pressure' mean double the airflow? For instance, if there was a marble in the middle of this 'box' just floating there, would more air pass around it in a certain time if the intake=exhaust method versus the 'hole method' described above?

 

[Cliff_J]

KingNothing - in the case of a computer where the restriction is going to hurt fan performance, a second fan assisting in overcoming that resistance by doubling the pressure difference will help. Since its not necessarily a linear relationship, it may result in slightly more than double the airflow since the fan performance is generally better at higher flow rates. The link from Russ has a lot of good information in that regard.

Blowers are sometimes called squirel cages because they look like the wheel you put in a gerbil cage so they can exercise.

 

[Danger]

Okay, you've got me totally confused now. Don't the blades of a fan act as a restrictor valve that limits airflow to the cfm rating of the fan? If so, how can you increase that limit by raising the pressure? If the airflow exceeds the speed that the fan is trying to pass it at, then the fan itself seems redundant.

 

[Artman]

Okay, you've got me totally confused now. Don't the blades of a fan act as a restrictor valve that limits airflow to the cfm rating of the fan? If so, how can you increase that limit by raising the pressure? If the airflow exceeds the speed that the fan is trying to pass it at, then the fan itself seems redundant.

Fan curves are a graph of flow and pressure. As the external static pressure the fan has to overcome increases, the flow decreases. So adding another fan in series to reduce the static pressure required to move the air, will allow the fans to both move a higher volume of air.

If you make both fans blow into the box with none blowing out, the static pressure acting against the fans will increase and instead of getting twice the amount of air into the box, you will get less than twice as much as both fans fall back on their curve to the higher external static pressure point on the curve.

 

[Danger]

Ohhh... Thanks, Artman.

 

[Artman]

Ohhh... Thanks, Artman.

I think the idea that the pressure inside increases confused a few people. All that does is to decrease the external static pressure required to remove the air, allowing more flow.

 

[russ_watters]

Arright, let me see if I can fix the mess I just made...(and add to Artman's explanation)

KingNothing - in the case of a computer where the restriction is going to hurt fan performance, a second fan assisting in overcoming that resistance by doubling the pressure difference will help. Since its not necessarily a linear relationship, it may result in slightly more than double the airflow since the fan performance is generally better at higher flow rates. The link from Russ has a lot of good information in that regard.

No. For fans in series, you get double the available pressure at the same flow rate. But since pressure drop across the case is a square function of velocity, you do not get double the flow. Figuring out exactly what flow you get is complicated because you don't really know the actual pressure drop across the case. If you could measure it, you could then match it to the fan curve or calculate the system curve. But to make matters worse, dc motors do not run at constant rpm, so you don't even know which fan curve to use. RPM is generally a function of CFM, and you actually get more rpm with less cfm, as air resistance for the blades decreases.

If you were to put the two fans in parallel, then they become capable of double the flow rate, but at the same pressure drop. But a lot of the same problems apply.

Now, 50cfm is a lot for such a small area, so at the very least, I can assure you there is no way to get 50cfm through a computer case with anything less than a bathroom exhaust fan.

KingNothing - stack effect is what you get when you confine convection with a vertical duct of some sort. The stack allows the air to accelerate in a coherent stream instead of just floating up in a bubble like in convection.

Anyway, the whole point of what I was saying is that when you are trying to put a lot of cfm through a small space, static pressure is what you have to worry about. Since a bladed fan can't generate much, all that'll happen is that you'll stall the fan and it won't give you much airflow.

Take the highest capacity 120mm bladed fan on that site I linked: It has a chart that shows 129cfm, and a static pressure of 8.1mm. But if you look at the graph, those are the endpoints and you don't get them at the same time. Picking a couple of points on the curve, you can either get 70cfm @ 3mm of static or just 26 at 5.2mm of static.

Now look at the 120mm blower: it gets 26cfm at about 7mm of static.

For comparison, 7mm is .27inches, which is about as low as you ever see a commercial fan (such as a bathroom exhaust fan) at. And fans like that get only 50-100cfm (depending on size).

 

[Artman]

Good explanation Russ.

 

[KingNothing]

Arright, let me see if I can fix the mess I just made...(and add to Artman's explanation) No. For fans in series, you get double the available pressure at the same flow rate. But since pressure drop across the case is a square function of velocity, you do not get double the flow. Figuring out exactly what flow you get is complicated because you don't really know the actual pressure drop across the case. If you could measure it, you could then match it to the fan curve or calculate the system curve. But to make matters worse, dc motors do not run at constant rpm, so you don't even know which fan curve to use. RPM is generally a function of CFM, and you actually get more rpm with less cfm, as air resistance for the blades decreases.

My DC Fans run at pretty constant speeds, if that could help my problem any. I'm aware that high CFMs are unrealistic inside of a case. I think what you are saying here is basically the result of how I model the airflow in my head.

I also have kinda figured out that I may be taking the wrong approach to this whole problem. CFM is simply cubic feet - but as for a radiator (using it in a broad sense, anything that radiates heat) giving off heat to the air, the air density matters too. 50 CFM of dense air will cool something faster than 50 CFM of low-density air at the same temperate, correct? Now this whole idea of which is better is just giving me a headache.

 

[FredGarvin]

I also have kinda figured out that I may be taking the wrong approach to this whole problem. CFM is simply cubic feet - but as for a radiator (using it in a broad sense, anything that radiates heat) giving off heat to the air, the air density matters too. 50 CFM of dense air will cool something faster than 50 CFM of low-density air at the same temperate, correct? Now this whole idea of which is better is just giving me a headache.

That is exactly the reason why all air delivery systems, i.e. compressors, fans, etc... are referenced back to a standard set of conditions of pressure and temperature (or at least should be). That way you have an apples to apples compare of what you are looking at. It drives me crazy when I see a product rated in CFM instead of SCFM.

 

[KingNothing]

That is exactly the reason why all air delivery systems, i.e. compressors, fans, etc... are referenced back to a standard set of conditions of pressure and temperature (or at least should be). That way you have an apples to apples compare of what you are looking at. It drives me crazy when I see a product rated in CFM instead of SCFM.

So essentially what I should be looking for is a configuration to pass the most molecules of air through the case.

I have two plans:
1. A case with two top exhaust fans and air inlets on the bottom for air to be sucked in.

2. A case with one exhuast fan on top, and one intake fan on the bottom.

Which arrangement would get more molecules to pass through? We can make up numbers, I'd just like at least a logical explanation.

 

[russ_watters]

While what Fred said is true and hugely important in some instances (I recently worked on an HVAC system in Chihuahua, MX, where the air is 15% less dense than at sea level and it was a big issue there), for computers you can ignore that. At the pressures we are talking about here, the density variations are insignificant. Concentrate on maximizing cfm.

Regarding computers specifically, I rather suspect the biggest airflow issue is with the grilles. It is, of course, case dependent, but many cases just have an array of small holes covering the place where you install your exhaust fan. Such an array has probably 25% free area, meaning if you have a fan with an area of 4 square inches, blowing it through that grille is like blowing through a 1 square inch hole. Improving such individual sources of pressure drop will go a long way toward improving flow.

The second big issue is where the flow is going. You need to find a way to get the coolest air to the front of the CPU and GPU cooling fans. The usual way is via brute force at the exhaust end, bringing so much air through the case that you end up with very little difference between temperatures inside and outside the case. Another way would be to supply air directly to those key areas. I solved a crappy case design once by mounting a 120mm fan on the side of the case, so it blew its air directly onto the cpu.

I'm actually a little curious about some of these issues, and depending on time and boredom, I may do some testing of fan performance. I searched without success for a review site that actually measured static pressure and cfm. I have some tools from work that I can use to relatively easily measure such things. First I need to buy that blower I linked, though...

Regarding your two ideas, I'd probably go with option 1. With a slightly negative pressure, you'll be pulling air through every gap in the computer case, pulling it across things like the hard drives. They don't require much cooling, but a little bit can mean the difference between a live and a dead one (I've had issues with them in an overcrowded case in the past). Negative pressures also mean lower intake velocities and as a result, less ability to pull in dust.