Acoustic chemometrics on liquid flow: Shift in the frequency spectra and its relationship to the physical properties of the liquid and the pipe

Prediction of chemical composition from passive acoustic measurements using multivariate regression (i.e. acoustic chemometrics) from liquid flows has been reported as a promising on-line measurement method. It is recently reported that the important predictive information could be related to frequency shift of peaks in the acoustic spectra. Little casual interpretation of the acoustic spectra is reported. The present study discusses the relations between chemometrically identified significant peak shifts, with prediction information in the acoustic spectra, to physical theory for vibration (sound) in pipe and liquid flows. Acoustic spectra were obtained for the pipe filled with the test liquids both for the situation with no flow and constant flow with a resulting pressure drop. The liquids were two-component mixtures of ethanol and water, and three-component mixtures of ethanol, sucrose and water. Experimental data were compared to the physical theory. Results indicate that some of the peaks in the acoustic spectra relate to bubble oscillation. The other peaks relate to shell (pipe) vibration. In the two-component mixtures measured at a constant flow 14 peaks were identified as important for the prediction of ethanol. Five of these peaks seem to be related to the bubble oscillations and nine peaks seem to agree with the shell vibration theory. The relations between experimental data and physical theory in the three-component mixtures of ethanol, sucrose and water were more complex. Seven peaks were identified as important for the prediction of ethanol. These peaks seem to agree with the bubble oscillation theory. Five peaks were identified as important for sucrose prediction. These peaks seem to relate to the shell (pipe) vibration. The prediction ability of the acoustic spectra for each of the constituents was reasonably good. The density of the liquid is probably the major factor, affecting the peak shift, given an otherwise fixed set-up.