Search

To identify the major sources of noise radiated from in-duct fluid machines, one needs to observe the source parameters such as pressure and velocity fields at the source plane. For this purpose, inverse estimation methods are generally used starting from measured pressures and a sound propagation model in the duct. In this paper, a technique is suggested for identifying the rotating noise source in a wide duct. For the detailed observation of sources with high spatial resolution, the evanescent wave is considered in the modeling based on the modal summation method. Also, the Doppler effect caused by rotation of the noise source is considered in the modeling. The validation experiment is conducted on the duct system excited by a rotating loudspeaker radiating a tonal sound in the absence of flow. The measured near-field pressure precisely shows spectral peaks at shifted frequencies due to the Doppler effect. The modal amplitude set related with the rotation of the loudspeaker is estimated to investigate the source parameters. The pressure in the near-fields very close to the source is regenerated by using the estimated modal amplitudes, and the maximum error is found to be less than -10 dB. The pressure and velocity fields at the source plane are restored by the estimated modal amplitudes, and the result clearly indicate the rotating loudspeaker as the main noise radiator within the rotating reference frame., QC 20200706

While an acoustic wave propagates along a duct, of which the wall is treated with either dissipative or a reactive material, the phase speed can be slowed down because of wave dispersion. It has been thought that such slow sound can be used for a novel control method to reduce the in-duct noise at low to medium frequencies generated from a fluid machinery system. In this work, the Herschel-Quincke tube (hereafter, H-Q tube), which exploits the path length difference of two parallel ducts, is modified to demonstrate the application potential of the slow sound. A test rig is designed to create the two different phase speeds by arranging the two parallel, equal-length ducts inside a main duct, one of them is hard-walled and the other one lined with a periodic array of resonators. This slow sound H-Q device is then modelled by both analytical and numerical methods assuming a plane wave incidence. Also, an experiment is conducted to measure the transmission loss. The result reveals a low frequency peak (TL-30 dB) in the range of 200-400 Hz, which occurs far below the lowest resonance of the resonator. At the original resonance frequency of 691 Hz, a small attenuation (TL~6 dB) is obtained due to the fact that one duct is subject to a high loss, and the other is without appreciable loss. The result clearly demonstrates the potential of applying slow sound device to overcome the spatial limitation of the classical H-Q tube., QC 20180319

The detailed information on the distribution of acoustic pressure and axial particle velocity at a source plane in a duct is useful for the low noise design of the source because it provides the position and strength of major noise sources. However, due to the limited spatial resolution of the conventional methods, the source behavior could not be observed in detail. In this work, a method to identify the pressure and axial velocity distribution at the source plane with high spatial resolution is suggested. Modal summation method including the evanescent waves is formulated, in which data can be provided from the near-field measurement. For the validation of the present method, the near-field array measurement in a duct excited by a blower is conducted. In the measurement of the nearfield and farfield acoustic pressures, the flush mounted microphone array is employed, and reference microphone technique using the cross correlation is adopted to suppress the turbulent flow noise. As a result, the measured near-field pressure is well restored with 26 evanescent modes, and the error between measured and restored pressure spectra is less than -20 dB in the Helmholtz number range of 0.57 < kR < 1.84. It is observed that the flush mounting measurement tends to over-estimate the pure radial modes, which results in the un- physical error in the predicted source parameters at the center of source plane. As a remedial ac-tion, pure radial modes are removed from the model that results in a successful reconstruction of the source distribution., QC 20150611