I had a new Coleman 99.5% gas furnace installed.
The return duct is a 25' 16" diameter metal straight shot from the
hallway to the furnace in the attic. Air flow is about 900 linear
feet/minute. That's over twice the airflow for my 37 year-old Lennox.
I get lots of furnace noise in the hallway. There's "wind" noise plus
what sounds like rumble from the blower. Noise level is 74dBC/60dBA
at ear level under the duct. The difference in the numbers suggests that
the problem is in the low frequency area. A spectrogram of the noise
confirms that the peak is at 100-200Hz. area.
I've got 9 source vents in the floor. Noise from them is tolerable.
Googling suggests that there should be at least three
right angle bends in the return duct to abate the noise.
I couldn't find any more detailed recommendations.
There will be straight segments between each right angle bend.
What's the optimal/practical length for each segment?
As the length of the segment approaches zero, it's contribution
to noise abatement will be reduced toward zero.
As the length of a segment increases, the marginal effect on noise
abatement decreases. Somewhere in the middle should be a sweet spot
for segment length. What should that segment length be???
How does that relate to the noise frequencies involved?
Would different lengths for each segment be better to avoid resonances
or spread the
losses over the frequency range of interest?
The contractor wants to insert 25' of flex duct and "serpentine"
it between the ends after cutting out a section of the straight pipe.
His idea of a right angle is to cut another hole in the side of
the return plenum. That's not an effective right angle at all.
How does inserting curves of flex duct compare with right-angle joints?
How would one circular loop compare? That would take minimal adjustments
to the existing pipe.
Discussion suggests that one of us (me and the contractor) doesn't
understand the optimum duct situation.
The contractor is not willing to discuss noise level goals.
He wants to modify the ducts for a fee, without making any
promises about solving the noise problem.