Dadodad, your interpretation of the Engineering Toolbox graph left me scratching my head, trying to understand the major discrepancy with my own experimental observations. I think I may have figured it out.
Here’s the thing: if you reduce the air flow rate through a given ducting system, the pressure drop will also be reduced. I’m certainly no fluid dynamics expert, but my recollection was that the pressure drop is more-or-less proportional to the square of the flow rate. Which is consistent with the power being related to the cube of fan speed, given that the mechanical output work of the fan is equal to the airflow rate times the pressure differential. I found a good reference just now that appears to confirm my recollection.
So let’s revisit your example, except using 800 Pascals as the baseline pressure, at 0.6 m3/s flow rate, just to make it easier to read the graph. At those baseline numbers, the graph shows a Sound Power Level of 96 dB.
Now let’s cut the flow rate in half, to a volume flow of 0.3 m3/s. The static pressure goes down as the square of that change, i.e. by a factor of four, which brings it all the way down to 200 Pascals. In that scenario, the graph shows a Sound Power Level of 81 dB.
I’m sure you’ll agree that reducing the fan noise from 96 dB to 81 dB is indeed a major reduction.
However, my “perceived” reduction in noise was even greater. I suspect this is because the quality of the sound also changes. My ears find high-frequency noise to be much more objectionable than lower-frequency noise. I have absolutely no objective data to back this up, mind you, but it stands to reason that running the fan at lower RPM will shift the noise-frequency spectrum downward. The Engineering Toolbox graph doesn’t address this difference at all, but presumably just includes the sound power contributions from all portions of the audio frequency spectrum.
In any case, I stand by my previous report that a 20% reduction in fan speed resulted in major noise reduction, at least for my particular dust collector and my particular pair of ears. When I have a bit more spare time, I’ll download a sound-meter app to my smartphone, and report back here with some more objective test data.