Probably a stupid question, but...

I've always wondered which is best.

Do you guys think it is smarter to have a fan blowing air onto something that is getting hot, or sucking up the air around the thing getting hot. I'm not having any kind of a problem, just curious.

There's no easy answer to that.

It depends on so many things - the airflow type (turbulent or laminar), the pressure - both ambient and the pressure change across the heatsink due to things like the Venturi effect, the velocity and how the velocity varies over the surface, the surface area and geometries...

There may be a general rule for which is better. I've never seen conclusive evidence either way. In the end the difference is probably usually negligible.

my idea would be that sucking hot air up would be better because then ther colder air would flow to where the hot air was and this would work somewhat as a convection oven just rather than heating we're talking about cooling

I have no real knowledge about cooling, but I find in general it's more efficient to pull the heat away than to try and cool it off. For example, if you think about it, you want your CPU to get colder, so the heat sink is designed to pull the heat as far away as possible, I want the air to be an extension of that. I'd also want some fans to bring the air in, but, as in many pc cases, I'd want them at the front, so they're not really blowing on the heat sink, just providing the air to be pulled out, create a nice strong flow, you know? Of course, if you can keep that intake air as cool as possible, that's always a plus too (aka avoid sucking the air past an array for 4 HDD's, which so many cases like to do).

Well that's one half of the argument. But, what if instead of relying on relatively cold air to fill the hot air's void on its own accord, I push cool air in which in turn pushes hot air out?

You more or less accomplish the same thing, which is to say that you're replacing air that has absorbed thermal energy from the heat source, but there's some subtle differences:

Your configuration more or less represents a vacuum. You are trying to pull air away from the heat source - if this was a closed system and no air would replenish what you've pulled away with a fan, leaving relative vacuum behind. With a complicated geometry like a heatsink what tries to have a lot of surface area, you get a lot more resistance to air flow - technically in gases you talk about conductivity instead of resistivity, but it describes the same kind of thing as in a electrical circuit. This conductivity places a limit on how fast you can evacuate air, but it also places a limit on how fast the air can be replaced. Furthermore, a change in pressure or number of gas molecules on the vacuum side affect the pump rate as well; in this case the (I think) expected pressure drop and concentration drop mean that the rate at which you're removing air is going to decrease. If the removal rate decreases, then with air being replaced at the same rate, you'll begin to build a higher pressure and concentration, and so on, until you reach a steady-state. Keep in mind that with the velocity around complex geometries and temperature differences... you're going to get some other changes in pressure and whatnot.

My configuration represents a pump, which is to say, I'm forcing more gas in. Again, there is some conductivity that limits how fast I can push gas in, and the rate at which I pump in depends on the pressure on each side of the pump, the temperature and the gas concentration as well. You'll get the same behavior - pumping causes a pressure and concentration change, which in turn affects the pump rate, and so on until you get to some steady state flow rate.

So, the main differences - it would seem like your method would be associated with some net pressure drop around the heat source, while mine might be associated with some pressure increase. Which is more conducive to thermal energy transfer? The answer, as with anything engineering, is that it depends... In a vacuum, heat transfer is poor. Why? Because heat is transferred by the gas molecules, and if the concentration is very low, then there aren't many molecules to transfer heat. At the same time, start increasing the pressure too much and you'll probably start encountering other effects.

Another big question is, how do each of our methods affect how the air flows over the heat source? This gets into the flow regimes I mentioned above, turbulent and laminar. Technically there's also a transition region and then molecular flow, but if you're in either of those two you've got problems already. Turbulent flow is characterized by eddy currents and disturbances; laminar flow is characterized by smooth slipstreams of air. Which you have depends on geometry, velocity, etc... which is better for heat transfer?


Just to make it clear - and it might already be clear to those of you with expertise in these areas - I'm no expert in thermodynamics or fluid dynamics. I'm just applying what I've been exposed to. And, just to reiterate... I think that in the majority of cases, the difference will be negligible.

Just what I hoped for, a controversy. :P

In theory, a fan pushing cold air to something hot will experience less heat stress itself and last longer than a fan pulling hot air away (and heating itself in the process). I don't have any evidence to support this, though.

I'm a believer in pressurising the case (blowing in the front and blowing from the side ducted to the cpu) and controlling where the air comes out (across the gpu and through the power supply).

Ben N1NP

I would go for...blowing cool air on a heatsink as apposed to pulling it away
With side off and 8" house fan blowing cool air in


Not to mention in the dead of summer and your sweating ur nuts off and theres a fan in the room...are you going to stand behind it ?...drawing heat from yourself
Or would you stand in front ???