Computational Modelling Methodologies and Experimental Verification of the Vibro-Acoustic Behaviour of a Section of Aircraft Fuselage

Thumbnail Image
Koukounian, Viken
Acoustics , Vibrations , Computational Modelling , Boundary Element Methods , Finite Element Methods , Experimental testing , Acoustic Testing , Transmission Loss , Acoustic Damping , Acoustic Distortion , Flanking
An aircraft is an example of sophisticated engineering requiring a high level of understanding, where advanced modelling techniques are used to improve design. This investigation identifies noise and vibration (N&V) as principal metrics. The considered source of N&V is the Turbulent Boundary Layer that causes the aircraft fuselage skin to vibrate, generating noise inside the cabin that dominates the sound field between 100Hz and 5kHz (frequency range of interest, FRoI). New computational methodologies (CM) are proposed to model and predict the vibro-acoustic behaviour of a section of an aircraft fuselage. The methodologies were adapted to incorporate more complex configurations, and materials. To verify these predictions, an atypical acoustic facility was commissioned to establish a new experimental methodology (EM). Several room qualification tests investigated the characteristics of the facility. The stringent criteria defined by these tests, and associated standards, were satisfied. The principal objective of the investigation was to develop a repeatable and reproducible strategy incorporating the complementing methodologies. The principal motivation reviewed the effects of varying thickness, and milled pocket size of the interior surfaces of the fuselage panels on Transmission Loss, TL. An associative formula was developed to account for the differences (damping, distortion, and flanking) between the methodologies. These corrections were all experimentally measured. Across the FRoI, the maximum TL error between the CM and EM for 0.04in, 0063in (milled), for a 0.063in, and 0.09in panels are 4dB, 4dB, 4dB, and 5dB, respectively; the error for the 0.063in and 0.063in (milled) panels with damping material are 9dB and 15dB, respectively. The TL error between the CM and EM across the entire frequency range (20Hz to 20kHz) for all mentioned panels vary as high as 20-25dB, however the frequency-averaged error for the 0.04in, 0.063in (milled), 0.063in, and 0.09in panels are 3dB, 3dB, 2dB, and 0dB, respectively; the frequency-averaged error for the 0.063in panel and 0.063in (milled) panels with damping material are 11dB and 17dB, respectively. In summary, thinner panels demonstrated greater attenuation per unit of weight (despite thicker panels demonstrating greater attenuation). Similarly, the application of damping material on the interior surface of the wall demonstrated greater attenuation but did not improve the attenuation per unit of weight.
External DOI