Good news. Some testing data on the Defender fan has arrived

Well we finally have some information from the boffins at University of Canterbury. Specifically a K. Shaw and Dr John Pearse.

The testing report can be downloaded from here.

The first thing I would like to say is that is appears to be a quieter machine than the Amarillo. This would not be difficult of course, but credit where it is due, any improvement is not to be sneezed at.

The second thing of note is that it appears that FMR has been publishing incorrect data on the machine in its advertising. In the latest brochure (here (~500kB)) they state that "At full operating rpm the new Defender fan is producing less than 50dba(L10) at 300mt". I could be nit picky and poke the borax at the advertising wienie's understanding of basic physical measurements (that should be 50dBA (L10) and 300m (there is no 't' in the metres abbreviation, clowns)). However, an astute reader will note that I would seldom stoop to such a thing. Instead I will quietly point out that no test shows that the Defender ever produced less than 50dBA (L10) at 300m. In fact the testing states that the only test carried out at 300m produced 52.0dBA (L10). Almost twice as much as claimed. (on the basis that a 3dB increase is a doubling of noise energy).

There are several other tests at various distances, (100m and 200m) but only one at 300m.

The last point and the most interesting is that the claim of "less than 50dba(L10) at 300mt" was most likely derived from a predicted value for noise from the fan. Not measured.

More interesting still was the method of prediction. You see the 300m test wasn't done in frosty, inversion layer conditions. Instead they were done in 11 degrees centigrade and a cross-wind. There is a prediction method that allows for compensating for differing climatic conditions when measuring noise. This is the CONCAWE method (1MB) (there are infact several different methods).

Sadly this method does not allow for the presence of an inversion layer with it's associated noise refraction properties and worst of all it is not designed for the noise from frost fans. Instead it is designed to be applied to petrochemical installations. Now I would be happy to say that the prediction methods for frost fan noise using the CONCAWE method were valid, but that has never been demonstrated, and quite frankly when the testing presumes to predict that noise from a frost fan will be quieter when an inversion is in place on a calm frosty night, I have my doubts.

By the way, what frequency ranges were being considered in the noise report? I couldn't see any stated, yet it appears to be a big deal when doing the CONCAWE calculations. I suppose the frequencies most likely to be associated with a frost fan?

I'm sorry Dr Pearse, some testing to validate the CONCAWE model against the noise characteristics of a frost fan in frost conditions would be required. Otherwise any predicted values are assumptions, not conclusions.