Computational modelling and experimental verification of the vibro-acoustic behavior of aircraft fuselage sections

The aerodynamics of an aircraft impose significant stresses upon its structure. The turbulent boundary layer (TBL) is a highly turbulent layer that forms along the fuselage skin inducing localized pressure fluctuations resulting its vibration, and in turn, the generation of noise inside the passenger cabin. During flight, the noise generated by the TBL dominates the sound field between 100 Hz and 5 kHz, to be regarded as the frequency range of interest. While the audible range is between 20 Hz and 20 kHz, human hearing and speech intelligibility is most sensitive between 250 Hz and 2 kHz. This investigation considers a BEM-FEM-BEM modelling technique to predict the vibro-acoustic response of the fuselage and an experimental methodology to verify the results (following ASTM and ANSI testing standards) by imitating the frequency profile of the TBL using an acoustic source. The research incited construction of an atypical acoustic testing facility, the development of DAQ software and post-processing techniques of test data. The principal quantity of interest is transmission loss. Four panels (0.04 in., 0.063 in. (milled pockets to 0.043 in.), 0.063 in., and 0.09 in. in thickness) were simulated and tested between 20 Hz and 20 kHz. Analysis of the results sought to determine the limitations of the computational methodology by observing divergence of the predictions from the results. Divergence was defined as a difference exceeding 10% (approximately 4 dB), which was observed beyond 8 kHz. The comparisons show the frequency-averaged errors between the proposed methodologies to be within 5 dB between 20 Hz and 20 kHz, and 3 dB between 100 Hz and 5 kHz. Variability in reproducibility of experimental results (same test specimen and between test specimens) is a significant challenge when determining transmission loss values. The experimental methodology proved successful in differentiating between the panels with confidence using at least six tests over a period of three years. The computational methodology was accurate in estimating the transmission loss and the general (frequency-dependent) response.