a (very little) bit of fan test data

[BillA]

there are a number of excellent fan info sites, Dave Smith's (http://www.amdmb.com/article-display.php?ArticleID=132), and Mike Turner's (Comair Rotron) (http://www.electronics-cooling.com/Resources/EC_Articles/MAY96/may96_01.htm) and another by dewitts for which I now get a 404
(if someone has a good link, please post)

nihli sent me some 80mm fans so I tested a pair of Sunon KD1208PTBX-6A fans to see if they could be stacked to any advantage

and the answer is . . . NO
all were measured in 6 in. duct with a Kurz thermal anemometer 12 dia.s downstream

1 fan = 42.5 cfm
2 parallel = 99.1 cfm
2 stacked = 23.0 cfm
2 stacked with 1 in. standoff = 16.8 cfm

when 2 were mounted on a Senfu rad, 55.8 cfm (vs. 46.0 cfm with the stock fans)

but one useful "tip" was gleaned:
as mounted on the Senfu, 2 sides of each fan had their discharge area blocked by a 1/4 in. "lip" of the sheetmetal case
after cutting this away, the thruput increased to 64.8 cfm
a 15% increase !

this proves that those round fan openings are inadaquite to deliver a fan's potential
the correct shape is a square with rounded corners
LOOK at the fan casing on it's discharge side
(and yes, it's a lot more work; but you'll have MORE output, and LESS noise)

be cool

[ButcherUK]

Interesting figures there Bill.
Odd that parallel seem to gain performance - 99.1 > 42.5*2. Maybe the mixed airflow patterns alter the pressure on the exhaust side of the fans?

[BillA]

a couple of different factors are at work here

the measurements described were all taken in an open duct
with only a nominal backpressure

later today I will run a single through the Senfu that I'm testing with (for comparison)

I understand the gist of the Papst article, again it is the backpressure that will determine the fan(s)' output
said backpressure being also quite dependant on how much air is being pushed

for example, look at fan panels for rack-mounted equipment, 9 axial fans running cheek to jowl
I bought such a panel with four 172mm thick-bodied Nidecs, which I then re-configured as two stacked pairs - but all four are pumping into a 6 in. duct !!

most radiators have an open discharge (and NOT into the case), hence fairly low backpressures at reasonable flow rates
- and this is why the rad article included a (comparative) graph of the different rads' flow rate vs. backpressure
(I'm thinking some aspects of that article are poorly understood.)

one NEEDS to use the fan mfgrs P-Q curves

it is a poor idea to try to use multiple low noise/slow speed fans, that's NOT what they were designed for

serious air movement requires thick bodied, higher speed fans that will still have some output at higher the backpressures that are a consequence of higher flow rates

be cool

[ButcherUK]

I'm aware of the backpressure issue, I have a 172mm thick bodied fan for my rad.
Just think it's curious that you get more than double the throughput with side-by-side fans, while I've seen other reports of double and less than double. In free air you'd expect double flow, so why 99 not 85 cfm?

[The Overclocker]

quite surprising, i would have expected a 25% increase buy cutting away the metal but this suggests that the fans spin faster when dealing with less air

[RoadWarrior]

Not so odd really when you think about it, that the parallel output was higher. Since the parallel mode stops the circulating churn flow on the sides of the fans next to each other. If all tests were done with the fans mounted to a flat plate that otherwise blocks the test duct there would be no circulation of flow around the edges of the fan. You'd use the correct shaped holes of course. Then parallel would probably be exactly double single flow rate.

Also it could have been some relationship between the area of active fan and the area of the duct. One fan was not enough to encourage a turbine/venturi type effect, two were.

Anyway. Experiment shows that stacking identical fans is a waste of time. Now if you had a pair of contra rotating fans....... :)

Road Warrior

[RoadWarrior]

Actually on a flat plate parallel might perform worse than double a single if right next to each other, since fans have a tendency to draw in air radially, so the radial flow between them would be restricted..... unless you spaced them a little....