One needs the data from 20 Hz to 4kHz for audible sound, some include the 8 kHz octave band. I thin one can spot the non industry authors by using a linear frequency scale and an overemphasis on unimpotant details.

Mufflers for fan noise needs to be efficient at the blade pass frequency and upwards. There tends to me a maximum in the performance just before the soundfield is over cutoff for plane waves, usually below 3 kHz. Thereafter the efficiency falls due to resonances an propagation of sound with crosswise components.

Most of the sound generating mechanisms in ducted systems typically fall in this frequency range .. also this range makes it amenabale to focus only on plane waves in common duct dimensions (ka

Your question has a very simple answer, right in line with the 2 previous comments: because as humans we perceive the infra sounds and sounds mostly in this band, and because they correspond to the dominant excitation (e.g. duct resonances excited by flow and by ventilator's blades). To give a closer look, frequencies must be discussed with regard to duct diameter versus wavelength (cf Anders right comments on cut-off frequencies) - 1m diameter duct in a public building is associated to 6xtimes lower frequencies than the 125mm diameter of the vent of your bathroom! But whatever the configuration, no ventilation-related problem would ever arise out of this global 1Hz-3-4kHz band... (the same for wind instruments, by the way, from the largest church organs to the smallest fife!)

The calculation for peak frequency is (rule of thumb) No. of fan blades times RPM this will give you the fan peak frequency within sound spectral analysis. This will give you an indication of the frequency range you are working in. the sound spectral analysis will provide a sound / performance of muffler (attenuator) to be able to design the absorptive and wall flanking potential of the muffler. I realise this is not the direct answer to the question, but hope it is an assist for practical solutions - I suggest the further reading of infrasound, low frequency noise and spectral analysis

TL is predicted with sine wave input in CAE softwares ?

Can we have broudband input white noise for TL measurement ?

the infrasound0.01 Hz to 20 Hz and low frequency elements say 20 Hz to 40Hz would not benefit from White noise broadband input as the results would be flawed. the 30Hz range is essentially the boundary between audible and sensory acceptance of sound. I can't see the rationale of the TL measurement for this particular question

Because the human can hear it.

mechanic parts has an eigenvalue lover then 200Hz (12.000 rpm) and this is normally enough for all design.

Over 3000Hz is only beep sound and it is not easy to excite.

More then 8000Hz is only rush, and for most of the People not perceptible.