Wave-resolving modelling and numerical solution of aircraft fuselages

Kurzfassung

Quiet aircraft passenger cabins contribute significantly to the well-being and health of billions of air travellers. During design, a reliable prediction of the sound pressure level is seen as crucial decision-making basis. The wave-resolving numerical simulation of sound pressure levels in the cabin is investigated within this thesis, which results in a comprehensive modelling of a typical aircraft fuselage, the efficient numerical solution and the assessment of the sound induced by a novel engine concept and the turbulent boundary layer. A major challenge is the complexity of aircraft models including structural, acoustic and poroelastic domains. For each relevant fuselage part, experiments are conducted and different vibroacoustic models are compared in order to choose a suitable modelling approach, repectively. The airframe, the insulation, the interior panels and the cabin are studied separately and merged into a full aircraft fuselage model. One important finding is on the glass wool insulation layer between the airframe and the interior panel, for which the need for the Biot model is shown. The modelling of structure-borne sound transmission within the glass fibres is necessary to take structural resonances within the double wall gap in the low frequency range into account. Another finding concerns the cabin domain containing seats and passengers - a global homogenised damping approach yields suitable results compared to a more detailed consideration of local surface impedances. As the resulting finite element model incorporates millions of degrees of freedom, efficient solving approaches are studied with regard to potentially introduced errors. The combination of frequency-adaptive meshes and the admissible weak mechanical coupling assumption for the cabin above 410 Hz decreases the computational time required by 87 % for a 4.3 m fuselage section, while the reference is a constant mesh with a strong mechanical coupling between all domains. In addition, the application of the iterative solver GMRES in combination with a suitable preconditioner (Block low rank LU) is presented. The usage in blocks according to the physical domains requires significantly less computational effort compared to a direct solver. Besides, it is shown that the full length aircraft fuselage can be replaced by shorter fuselage sections without introducing significant errors above 410 Hz. The aircraft fuselage model in combination with efficient solving approaches is applied for two analyses. Firstly, the available acoustic excitations of two jet engines (bypass-ratio 17 and 5) are compared. The acoustic footprint of the novel engine yields significantly lower sound pressure levels in the cabin, which results in a positive assessment of the new technology. Secondly, the acoustic excitations beneath the turbulent boundary layer are modelled by a generic superposition of plane waves resulting in a deterministic snapshot of the random loading. It is demonstrated that the turbulent boundary layer generates cabin sound pressure levels lying between those of the two engine generations. Based on these results, a more distinct presence of the turbulent boundary layer acoustics within the cabin can be assumed for future aircraft generations. The assessment of complex excitations by applying the acoustic footprint directly shows a great advantage of wave-resolving models. The developed recommendations on aircraft fuselage modelling and the derived efficient solving approaches state the major scientific contribution of this thesis, which is transferable to different aircraft and even other mobility vehicles. The obtained results represent a decisive contribution towards the development of quieter passenger cabins in future aircraft.