Natural Science Forum / Physics / Acoustics / June 2004

on Mudd wrote:

 Thank you for helping put my mind at ease a bit on this. As far as i
 am aware the coil has one row and the ductwork is all rectangular
 galv. steel nominally 450 x 350. I am happy with the discharge
 velocity and the calculation for the room characteristics.

 It would be useful to have any data or prediction formulae for these
 types of coils in an airstream.

I have been working with flow generated noise "in general", which is the
perspective of the reply.
I like to separate different levels of estimates and broad band noise
from narrow band. Also the
separation between plane wave and higher modes is often done as you can
see below. If I remember
corretly all the broad band estimations include downstream sound power.
The upstream is about the
same. The flow velocities commonly considered are usully from 10 m/s up.

1.  "Rules of thumb" which usually take form like Lw = 50...60logV+
10logA+C1(+C2), where V is
flow velocity in the structure (either geometric mean or "real" max), A
is (usually) open (flow) area, C1
structure dependent "constant" and C2 "an addtional" constant. SI units.
This kind of approach is
commonly used grilles and such, haven't seen for batteries. Approach is
common in "engineering level"
books and usually (allways?) includes broadband noise. For example take
a look at Woods Practical
Guide to Noise Control and/or Handbook of Noise and Vibration Control
and/or Beranek, Noise and
Vibration Control and/or Beranek&Ver, Noise and vibration control
engineer and/or Munjal,
Acoustics of ducts and mufflers..
My first guess would take the "constant" from grilles or resistive
silencers with the same open area
percentage.
At least some of the above books include octave spectra (but don't
necessirily separate the modes, so
be careful).
Estimates for narrow/broad band conditions (Reynolds number) are
presented in more detailed
acoustic studies or in many flow mechanics/dynamics books (for batteries
take a look for example
Blevins Flow-Induced vibration).
2. Pressure drop (or equivalent) coefficient (pd). The basic formula is
as above with the addition of
term 30logpd. From the above books Beraneks introduce this approach as
well.
3. Total or static pressure drop or drag of the structure. These
appraches also include only broad band
noise, but separate noise generated on plane wave region and higher
modes. Articles on the approahes
are written (among others) by Gordon, Nelson&Morfey, Oldham&Ukpoho,
Oldham&Waddington,
Waddington&Oldham in (at least) JASA and JSV since 1960's.
4. Obviously CFD can be applied for pressure drop calculations and CAA
for acoustic estimates. In
case of narrow band noise CAA is "a fashion", so the number of
publications is huge.
The narrow band noise (as usually considered in duct and pipe systems)
is based on (local)
laminar-turbulent transition and is therefor sensitive for the flow
conditions and physical structures
(both details and system level). In case you wanna understand narrowband
noise you might wanna
begin with Blevins and move quickly to Rockwell&Naudascher,
Selfsustaining oscillations of flow past
cavities, Journal of Fluids Engineering, vol. 100 (June 1978), pp
152-165. And after that You can
spend rest of the life on the subject:).

The really rough conclusion is that broadband noise is dipole-like
beeing V^4...6 and Dptot^3 and
Dpst^2 (total and static pressures) dependent. The levels can be
estimated pretty well, say -+ 3...10
dB also on octave or 1/3 octave bands. The basic frequency of the narrow
band noise can be
estimated easily with resonable accuracy (say +-10...20%), the "actual"
frequencies and  levels are
much more difficult but possible.

Hope this satisfies you (at least for a while).

BR

ari