A parametric uncertainty analysis for impedance eduction based on Prony's method

摘要

Impedance eduction techniques are commonly employed to obtain the acoustic impedance of liner test samples in the presence of flow. However, there is little information available in the literature regarding sources of uncertainty in the measurements, and consequently in the impedance results. This paper investigates the main sources of uncertainty in impedance eduction techniques based on Prony-like algorithms. The Monte Carlo Method is used to conduct a parametric uncertainty analysis for each input variable in a direct method for impedance eduction. The results show that uncertainty on educed impedance depends primarily on liner attenuation and a better accuracy is achieved when the downstream source is used. The parametric study shows that the critical variables are the acoustic pressure and the microphone separation distance. Results suggest that is better to increases the distance between consecutive microphones then changing the number of microphones. Also, the Kumaresan-Tufts algorithm achieves better accuracy than original Prony's method. Results obtained with the infinite duct study suggests that Ingard-Myers boundary condition is biased with dependency with flow direction. The Monte Carlo Method is used to evaluate uncertainty levels on NASA benchmark data and the results corroborate the infinite duct analysis.