Suck or Blow?! 2

[stuff]

Gents,
I'm designing an electronics package and am now looking at the unit's thermal design issues.

I need to cool a box of optics in the unit that operates at 40-45 deg C, with 200W dissipated by the lamp.

I am selecting a tacho radial blower with 16cfm flowrate (overspec for now) but can't settle in my head whether it would be better to suck or blow air through.

It's a straight path from back to front. I'd like to reduce noise. My understanding is that sucking air is better as the pressure reduction helps in a heated enclosure.

Thank you in advance for any assistance

 

[Metalguy]

I think the more air in the box the better, so blowing seems better to me. But most important is the airstream movement itself.

 

[jsteverm]

The fan is a constant volume device and the cooling is a function of the mass of the cooling fluid. If you suck then you are putting hotter air through the fan and there will be less mass to move the heat. This means blowing is better.

 

[TheTick]

For what it's worth:

Most refrigerators I have seen (and repaired) suck air across the coils. The last thing the air does before it exits the shroud is cool the motor.

On justice and on friendship, there is no price, but there are established credit limits.

 

[pjhype]

I've always felt that sucking the air out of a box is better than blowing the air in. When you blow the air in, you must carefully balance the incoming air with the outgoing air in order to avoid turbulant airflow within the box that can be detrimental to cooling. Ideally, you want to create laminar flow across your components in the box, and this is more easily accomplished by placing the fan on the outlet side.

One disadvantage to this method, however, is that if you have "dirty" air on the outside, the negative pressure will want to suck the dirty air in and possibly contaminate your optics. If this is the case, you will need to provide filters on the inlet side. If this is not possible, you are stuck with blowing clean conditioned air into the box and you must then be very careful in how you handle the airflow. There are a variety of CFD codes that can be used to predict this.

pj

 

[drawoh]

A small fan or blower maintains a pressure of as much as an inch of water. This compares with an atmospheric pressure of something like 37 feet of water.

Your pressure ratio is approximately 1. It does not matter whether you blow or suck. Ask yourself where cool air is available, where you want the warm air dumped, and where you want the coolest air inside the enclosure.

JHG

 

[stuff]

Thanks for the input, this is useful. The 'dirty air' consideration is important but we have EM sealing and 5 micron dust filters planned so perhaps I shouldn't worry so much about that one.

Stuff

 

[owlscreech]

Suction systems are more effective in general compartment cooling. If you plan to use the same air flow to cool other components then suction is good because of the issues brought up by pjhype. Also the fan motor add heat to the flow. With suction the heat goes away. This also means that with suction you are allowed larger motors and greater flow.

But no matter how you do it, remember to bring the air in where you want to cool, not after it has passed over other components. You would be amazed what the "experts" do sometime. You might need internal ducting. Either suction or blowing works. My preference is still suction.

 

[Ralph2]

I would disagree with <quote>
jsteverm (Mechanical) Apr 29, 2003
The fan is a constant volume device and the cooling is a function of the mass of the cooling fluid. If you suck then you are putting hotter air through the fan and there will be less mass to move the heat. This means blowing is better.<end quote>
Sucking would remove the heat that is generated by the fan where as blowing would add that heat.
But, that is just my opinion and I have been wrong before...
Good luck
Ralph

 

[DLANDISSR]

STUFF:

Whether to suck or blow this is the question! Ha!

Whether sucking or blowing cools better? Their both about the same if you just want to cool the enclosure and not a specific item within the enclosure.

If you wish to cool generally sucking may be better because of adding the fan heat. On the othe hand the heat from the enclosure may heat the air (which lastly cools the motor) to hot to effectively cool the motor.

Calculate the temperaure rise both ways and see which gives the best results for your application. Which is more important adding the motor heat to the enclosure or over heating the motor.

If you want to cool a specific item blowing is better because you can direct the flow specifically. When sucking it is harder to direct the flow specifically and not always possible.

I'll stop now, lest some think I'm a blow hard or that my answer sucks. Ha!

Sorry about the humor!? This just struck me that way!

Hope this helps!

 

[automatic2]

Sucking is almost always better when you consider two things;
- air inlets in the enclosure can be strategically placed to provide most efficient inlet streams. Especially effective for multiple heat generating locations.
- your fan is generally more effective at moving ambient air (within the enclosure) which is typically the sum of the smaller cooling streams. Warmer air from ambient will have less cooling capacity if you blew it toward heat locations.

Of course baffling can overcome most differences, and is most usefull when temperature buffering is required to prevent injury caused by high temp. differential in applications such as lamp cooling.

 

[alpharam]

I generally agree with sucking the air as more efficient cooling practice. Blowing adds the heat of the heat motor (heating the fan itself is usually less important). When sucking, turbulence of the airflow is avoided, which is very important once you wish to cool the devices effectively (turbulence is inevitable when blowing).
I would like to bring another aspect to your attention though. Sometimes fire hazard demands can be decisive. Under some standards, you may not allow an accidental fire (if occures) to be blown outside. Therefore placing the fans is very important. Do not use flammable components or wires in the closest proximity to the fans (if applicable).

 

[stuff]

Thanks for everyone who has answered this thread so far. The answer is obviously product-dependent. Whilst there are parts withing this device that would require spot cooling I have confidence that a careful placement of inlet vents would suffice.

Reducing the convection turbulence is a good point. I guess this also helps overall to reduce the device noise.

We're going to run tests over the next few weeks and from that we can optimise vent placement and fan ratings.

Stuff

PS I asked my girlfriend the same question and got a completely different answer!

 

[borjame]

All the talk concerning air heat up is not really an issue, pumping 16 cfm will raise the temperature of the air almost zero degrees.

approx. energy to pump air = .5*(density*vol)*deltaV^2
16cfm = .267 cfs
density = .076 lbm/ft^3

air specific heat = .17 BTU/lbm/degree R (I'm making some constant volume, sea level, room temp, etc assumptions)

approx energy = 1.018E-2 lbm-ft^2/s^2 *(lbf-s^2/32.2 lbm-ft)*(BTU/778 lbm-ft)

4E-7 BTUs added to the air which is approx zero degree rise

Now that air heat up isn't an issue, you need to decide whether you are you trying to cool a specific component, a group of components, or the enclosure.

Specific component: BLOW - blowing on a component will raise the local h through impingement (unless the componeny has fins, which you want even flow over ALL the fins)

Group of components: depends on the shape, any intro to heat transfer text will answer this one

Enclosure: SUCK - removing the hot air will allow the buoyancy effects to be increased by increasing the delta T between the surface of ALL of the components and ambient in the enclosure. 16 cfm isn't a lot of air so I'm not sure if laminar vs turbulence effects are important. I think if you run the Reynolds numbers

Re = (air density*air speed*representative lenght)/ air viscosoity

I think your Re# will be under 1000 and you will be nowhere near the transition to turbulent flow. Let's not get nickpicky about flow around wires which will have higher Re#'s.