The sound pressure distribution was solved for each frequency.
The sound pressure distribution at the frequency of 770 Hz is
shown in Fig. 3 as an example. The excitation of the mode shape
w10 can be noted from the sound pressure variation.
At each frequency step, the cross-spectra between the measurement points selected according to Section 3 were averaged using
Eq. (19). The results were converted to 1/3 octave bands of centre
frequencies of 630–6300 Hz. In non-dimensional Helmholtz-scale
the corresponding 1/3 octave band centre frequencies are
He
c ¼ 1 :5 to Hec ¼ 14 :8. Using this non-dimensional form, the
results are usable at other speeds of sounds (i.e. temperatures)
and duct sizes. For clarity, the piecewise sound power weighting
factors are presented as a function of the limiting Helmholtz numbers. The simulated 1/3 octave frequency band acoustic power
weighting factors are presented in Fig. 4. Also the experimentally
determined weighting factors for the monopole excitation from
[15] are presented in Fig. 4. The simulated weighting factors with
the corresponding standard deviations rxy are listed in Table 2
for the studied 1/3 octave bands, i.e. Helmholtz-scale center frequencies Hec. For clarity, also the lower and upper 1/3 octave band
frequency limits Hel and Heu are listed in the table.
The standard deviations of the five studied microphone configurations compared to the reference as described in Section 3 are
listed in Table 3. This means that when one of these
three-microphone configurations are used, the weighting factor is
obtained from Table 2 and the total standard deviation is estimated
as the sum of the values in Tables 2 and 3.
As can be seen from Fig. 4, the sound power weighting factor is
approximately zero in the plane wave range. This is reasonable
since the plane wave in-duct acoustic mode shapes equals unity,
i.e., there are no modal cross-terms. The experimentally determined weighting factor deviates from zero due to inaccuracies in
the measurements. Note that the sound power weighting factors
are used in Eq. (20), which is only valid in the non-plane wave frequency range, that is Hec > 1 :5. In the plane wave frequency range,
the downstream and upstream propagating waves should be separated and the in-duct acoustic power derived using Eq. (13). In the
mid frequency range there is some deviation as expected between
the experimental and simulated data. This is of course due to that
the experiments are done for a certain monopole configuration,
while the simulations are an average over a number of
configurations.
At higher frequencies, the sound power weighting factors
derived with the cross-spectra are converging to 6 dB value. As
can be noted from Fig. 4, a similar trend was achieved experimentally in [15]. The sound pressure at the duct wall is related to the
sound pressure averaged over the cross-section. According to
Joseph et al. [26], theoretically this relationship tends to two, that
is 3 dB. In the semi-diffuse field, the acoustic energy at a point in
the duct arrives equally from all angles over a hemi-sphere. This
also leads to a power weighting of 3 dB. In that sense, the convergence to the value of 6 dB is reasonable.
Fig. 2. Acoustic pressure measurement points used to estimate the in-duct acoustic
power. The distances between the sections are chosen to cover the plane wave
range in the classical two-microphone measurements [22]. The spiral microphone
configurations can be specified by defining the relative angular twist as the angle
between the two neighboring measurement points seen from the axial direction.
From the studied microphone configurations, the one with relative angular twist of
90 is shown with a solid black line.
Fig. 3. Sound pressure distribution at the frequency of 770 Hz. That corresponds to
the Helmholtz frequency of 1.8. The colors in the figure describe magnitude of the
sound pressure; ranging from higher sound pressure (red) to the lower sound
pressure (blue). The excitation of the mode shape w10 can be noted from the sound
pressure variation. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)