Application of cepstral signal processing techniques and a guard tube to the in-situ measurement of the normal incidence sound absorption coefficient

Groll, Neil

Application of cepstral signal processing techniques and a guard tube to the in-situ measurement of the normal incidence sound absorption coefficient

Thesis / Dissertation

1996

Abstract

The work described in this thesis is directed towards the development of a technique for the in situ measurement of the normal incidence sound power absorption coefficient of homogenous materials. The work stemmed from an International Standards Organisation (ISO) working group, ISO/TC 43/SCJ WG38, tasked to formulate a new international standard: " Procedure for measuring sound absorption properties of road surfaces: in-situ method'". The working group identified the need for the development of an easily transportable apparatus which would enable a single, non-skilled operator to rapidly measure the sound absorption coefficient of road surfaces, over a broad frequency band. The technique utilises a single stationary microphone housed inside a measuring tube, which is surrounded by a second tube of greater diameter, called the Guard Tube. Both tubes are equally insonified at one end and terminated by the material surface at the other. The source signal used was a swept sine signal which was modified using an inverse filtering process to extend the frequency response to zero hertz. The technique then employs cepstral signal processing to separate out impulse responses due to successive reflections from the material and that of the direct sound from the loudspeaker The direct and reflected signals can be measured at a single microphone position and are allowed to overlap in time The impulse response of the material termination is then directly extracted from the cepstral time history. Fourier transformation of this windowed data produces the reflection coefficient. After subtracting the squared modulus of the reflection coefficient from 1, the sound power absorption coefficient is obtained. Particular attention was given to the effects imposed by the loudspeaker on the signal processing technique, which was found to result in the corruption of the low frequency end of the absorption coefficient. Measurements performed on material sample cut-outs which were sealed in a sample holder at the end of the measuring tube, showed good correlation with the standardised Standing Wave Tube method, in the required frequency range of 200Hz and 2000Hz It is also shown that this frequency range can be extended 10 4000Hz. To be able to conduct a non-destructive measurement of the material surface. use was made of the outer Guard Tube in an attempt to offset pressure differences inside and around the inner measuring tube which arise in the absence of the sealing of the material at the end of a tube Measurement errors, particularly at the low frequency end of the absorption coefficient were caused by radially outward pressure leakage through the material, beyond the boundary of the Guard Tube. The thesis presents a model for quantifying this pressure leakage, which is dependent on the radius of the Guard Tube as well as the parameters of the material The model predicts that for a constant Guard Tube radius, there is, in general, an apparent increase in the low frequency plane wave absorption for a decrease in the specific flow resistance of the material. Measurements with the Guard Tube were then conducted on selected homogenous materials. Owing to the non-availability of porous asphalt surfaces and bored core specimens, the investigation was limited to more common acoustic materials. Particular attention was given to Glass Fibre materials which were easy to characterise in terms of their material parameters. Variations in measurement accuracy as a function of the material parameters correlated well with the theoretical model. The findings included a decrease in the lower measuring frequency limit for an increase in the specific flow resistance of the Glass Fibre materials. The frequency range for which the absorption coefficient results obtained with the Guard Tube were within 0.05 of the results obtained for sealed sample cut-outs ranged from a lower third octave frequency limit of 315Hz to 1000Hz, to an upper limit of 2000Hz. However, for relatively thin materials with a high specific flow resistance, the results were found to be accurate down to the required 200Hz. This thesis forms part of ongoing research and provides a foundation for continued work on the application of the Guard Tube to the in-situ measurement of sound power absorption coefficients of homogenous, porous materials.

Keywords

Electrical Engineering

Reference:

Collections

Masters

Full item page