regarding my own experiences with arctic 3 fans in my silentium t2 case i don't think the cfm measurements in the test are correct...
i also don't think arctic cooling states an incorrect cfm-rating of 28 cfm at their website. the company seems realistic on other matters....
using the foam-stand must have had some detrimental effect on the output of the arctic 3 fan in the test.
I did this particular cfm measurement and was also curious about it, so I set up this test jig and tried more experimentation.
I put the anemometer right up against the exhaust side, a half inch away, and an inch away. I held the fan up in free air, in the foam stand, and at the edge of a 4" tube so the intake side was shrouded. I also tried a 1" ring of cardboard around the perimeter of the blades, to box it in like a normal axial fan.
I also compared the feel of the airflow against my hand and my face from a few inches away -- with some other fans, the Panaflo 80L, the Nexus 80. All this at 12V.
CFM measurements on the foam stand or in free air was exactly the same as before -- a little above 12cfm. It did get higher in the 4" tube, with the anemometer held about 1" away; the highest reading was around 15cfm.
By feel, it blew a touch less air than the Nexus 80 (22cfm measured), which blew less than the Panaflo 80L (30cfm measured).
1) It does measure lower than expected, in free air or in our jig. The open frame may be a cause; I just don't know.
2) The fact that the cfm improved with the tube is suggestive.... but again, it's hard to interpret.
3) If our other cfm measurements are anywhere close to accurate, then I'd say there's no way I'd accept AC's 28cfm spec.
Our measured cfm seems too low for the AC, but we're not sure how to get a higher/more valid reading. We may try a proper jig again just for the heck of it...
Also, I would not be 100% positive that the fans in your case are the same as the fans we tested. AC have already gone though many generations of similar looking fans.