A Coupled Deterministic-Stochastic Approach for the Effects of Non Structural Components in Cabin Vibrations of Large Aircraft
Résumé
Passenger comfort is progressively gaining attention as one of the top requirements to ensure competitiveness amongst aircraft manufacturers; means for effective prediction of vibration levels in the cabin are required. Measured aircraft data indicates that the typical frequencies of rotation of the engine fan, from where small imbalances can excite the airframe, are located in a band where the fuselage response has attained common characteristics of the mid frequency regime, such as high modal density and modal merger. However, when only the load carrying structure is considered the response is driven by sparse distinguishable modes as is usual in the low frequencies. It is therefore concluded that the myriad of Non-Structural Components (NSC) inside the fuselage, such as systems, cables and others, play a paramount role in the global dynamic response of the craft already from surprisingly low frequencies and thus a different idealization for these items other than the classical lumped mass in the global FEM used for low frequencies is necessary to gain prediction capabilities at higher frequencies. Due to their large number and spatial distribution on the fuselage, cables are a particularly interesting category of NSC to consider the viability of new modelling techniques. The importance of cables in the global dynamics of aeronautical and space structures in literature is accepted but not yet representable through a numerical model even if notable efforts have been made in the field mostly through deterministic processes. In the proposed presentation it will be explained how through the available evidence a statistical formulation for NSC is deemed more promising for this particular problem. Inspired by the idea underlying the theory of fuzzy structures introduced by Soize in the mid 80’s, a numerical model based on the coupling in the frequency domain between a deterministic structure and a set of stochastically defined impedances will be presented. The model is defined in such a way to depend on a reduced number of parameters. It will be as well introduced an experimental set up which aims to validate the statistic treatment of the problem and to provide to an empiric database so far inexistent to the knowledge of the author.