ok, time for some education again, read up! older article, but 99% relavent
Fans 101
The humble fan is the most important cooling component in most PCs. Without it, there'd be little to no air flow through the case. Fans can blow air into the computer, suck air out, or just move air around inside. The fan on your CPU heatsink is a perfect example of the last kind; its sole purpose is to improve the effectiveness of the heatsink by increasing the airflow over it. These internal fans do nothing to lower the temperature of the air inside the computer - in fact, they slightly raise it - but they make that air do more work.
This is only any use, though, if the air is constantly being replaced with cooler outside air. As the air flowing over a heatsink approaches the temperature of the heatsink, its cooling efficiency drops. When the temperatures are the same, no cooling happens at all - you can blow all the 50 degree Centigrade air you like over a 50 degree Centigrade heatsink and nothing's going to happen.
Standard power supply fans suck air out of the computer and create the familiar warm breeze out of the back of your PC. Sucker fans are not what you want next to your hot components, because the air they move has already passed by other warm bits of computer, and is thus already warmer. A blower fan, on the other hand, delivers air from outside the box straight to whatever it's blowing on, and thus does more cooling for a given size.
In the official ATX specification, it says that the power supply fan should blow air into the case, towards the motherboard. Since this guarantees a steady flow of pre-heated air from the nice toasty PSU, many ATX case manufacturers ignore the spec and use power supplies with the traditional blow-out fan.
Many cases also have provision for a second fan, usually (in tower configuration cases) at the bottom front of the case. If you've got a dinky little mini-tower case with no extra fan mount spot and your computer has heat problems, get a better case. Full stop. If you've got a name brand PC and you can't change the case, I feel for you.
Front-of-case fans in most situations have a pretty hard time of it. They're usually trying to suck air through a fairly narrow vent in the plastic face plate and through a more-steel-than-holes grille in the front panel. Classier cases often have a dust filter as well, which lets the air through fairly well when clean but rapidly clogs to the approximate permeability of a blanket. Add to that the fact that the plastic cages that hold front-of-case fans practically never actually seal the fan against the front panel but instead leave it a half-centimetre or so into the case, so lots of air can leak around the sides, and it's amazing that most front-of-case fans do anything at all.
Check out this cruft, from the front grille of my old case after maybe a year of service. Filter? We don' need no steenkin' filter!
You can improve the situation somewhat by sealing around your front fan with cloth-backed Gaffa (or "duct") tape, like so:
If you're feeling more adventurous but don't want to go the whole hog and fit a bigger fan, you can also cut out the metal in the front fan grille and chop a couple of extra vent holes in the plastic front panel.
If two case fans are close to each other, they should both blow or both suck. Otherwise there'll be a storm of cooling air near the fans and not enough air movement elsewhere. If two fans are located away from each other, though, one should blow and one should suck, or they'll be trying vainly to change the air pressure inside the case and will lose a lot of efficiency.
A fan will deliver its maximum airflow when it's hanging in space - the so-called "free air" situation. It will deliver its minimum airflow - zero - when it's blowing into a sealed box. In the sealed box situation, all the fan can do is move a bit of air into the box, increasing the air pressure inside, and keep it there. These two kinds of fan behaviour are governed by the two fan performance statistics - pressure and airflow.
All computer fans are roughly the same shape, so their pressure and airflow statistics are roughly proportional to their size and thickness. Different kinds of air moving apparatus have different pressure to flow ratios - a "blower", for instance, which uses an enclosed impeller to push air out of a port, has a higher pressure rating but lower airflow than a fan of equal power. A hair dryer is a good example of a blower. It's also possible to buy small 12 volt blowers which can be used in computer applications, mounted inside the case to direct a well-aimed stream of air at something that needs cooling, like a CPU.
A computer case is neither a free air nor a sealed box situation. A really well ventilated case looks to the fan pretty much like the free air situation; the more tightly buttoned up and packed with cables the case is, the more like a sealed box it behaves. The lousy mounting locations of many case fans don't really make a lot of difference to the behaviour of the box, but they reduce the effectiveness of the fan; a poorly mounted fan behaves like a less powerful one.
Arranging fans physically in parallel - next to each other, blowing in the same direction - gives twice the airflow, but only in the free air situation. The closer the thing they're blowing into is to a sealed box - the higher the "system resistance" - the smaller the parallel fan improvement becomes. Any number of identical parallel fans blowing into a sealed box will do no more than a single fan.
Arranging fans in series - stacked on top of each other, so one blows straight into the other - gives the opposite situation. There's no improvement in free-air performance, but double the pressure increase in a sealed box, and better and better performance compared to parallel fans as the system resistance gets higher and higher. Incidentally, this means that those goofy monster processor heatsinks with layered low-profile fans are a waste of money, since fans strapped onto a heatsink are very nearly in the free-air situation, and putting two layers of them there does close to nothing to the airflow.
So series fans can pump more air into a box with poor through-flow, but parallel fans will do better in a box with high through-flow. When you arrange one fan blowing into a computer and another sucking out, you're basically making a poorly sealed serial fan arrangement. The PC box leaks all over the place, so not all of the air the blowing fan puts into it exits through the sucking fan, but the arrangement is much more like serial fans than like parallel ones. This is good; it keeps the airflow up even in the card-and-cable-packed mess that is many PCs.
The exact fan arrangement to use therefore varies with the box, if you?re going for maximum efficiency. The easy way around the problem is simply to go for overkill, mount much more powerful fans than you need, and have done with it. That's certainly what I settled on. As long as the fans aren't blowing components out of the back of the computer, the problem is solved. The problems with this solution are that big fans are more expensive, computer power supplies can be overtaxed by multiple fans (a problem you can solve by using a mains powered fan), and big fans make more noise than many users will tolerate.
12 volt fans with a three-pin connector plug into the special three-pin headers on recent motherboards and can thus report their rotation speed - if the speed drops, an alarm can sound. Some fans have the three pin plug but only two wires; they'll run from a motherboard fan connection, but won't report their speed.
In the near future, we should be seeing PCs with efficient ducted cooling built in. Ducts allow the cooling air to go where it's needed, and some fanatical overclockers already use duct arrangements of varying degrees of gimcrackness to keep their sweating and terrified components cool.
http://www.dansdata.com/hx45fan.htm