Bir ağır kamyon kabininde oluşan yapısal kaynaklı gürültü seviyesinin deneysel ve nümerik modeller ile tahmin edilmesi ve FRF tabanlı alt sistem analizi için yeni bir düzenlileştirme metodolojisi geliştirilmesi

Automotive manufacturers invest a lot of money and time to enhance the vibro-acoustic performance of their products. In a complex dynamic system such as a heavy duty truck cabin, the vibro-acoustic enhancement and redesign effort may be very difficult and time-consuming without prior knowledge about the noise contributors. This study proposes a design methodology that employs high fidelity computational models together with experimental methods to predict the vibro-acoustic performance of a heavy duty truck cabin. The analysis tools and the methodology presented provide a systematic and quantitative way to investigate the vibro-acoustic performance of such systems and identify the critical components of the system that degrade the performance. The noise inside the vehicle cabin originates from various sources and travels through many pathways. The first step of sound quality refinement is to find the pathways and corresponding operational internal forces. For that reason, acceleration responses and frequency response functions (FRFs) are measured on a prototype truck to determine the excitation forces while engine is running in operational conditions. Once these internal forces are identified using the experimental force identification (FI) technique, they can be utilized to predict the total sound pressure level inside the cabin and also perform the panel acoustic contribution analysis (PACA) to determine the most problematic panel of the cabin. An uncoupled vibro-acoustic finite element model (FEM) is used and validated with experimental measurements to predict the sound pressure level inside the cabin. When the most noise radiating panel is identified, it can be redesigned to improve the sound pressure level inside the cabin. Second half of this thesis focuses on response prediction with FRF-based Substructuring (FBS) technique where the FRFs of the truck cabin obtained from a numerical model is coupled with the experimentally measured FRFs of the truck chassis. The influence of each substructure modification, on the chassis and/or cabin, can be easily assessed by reformulating the FBS equations. However, the coupling procedure of the substructures involves an inverse problem which requires regularization techniques since the matrix being inverted can be ill-conditioned. The inverted FRF matrix is called the Kernel matrix. Accuracy of the predictions made with this technique is highly dependent on the inversion of the Kernel matrix and in general, singular value decomposition is used for the inversion procedure. In this thesis, a novel methodology is adapted for lowering the condition number of the Kernel matrix by selective omission of the cross-coupling terms such that the matrix inversion procedure can be improved. Response predictions obtained by the proposed regularization method are compared with the experimentally measured responses and also with those obtained through Moore-Penrose pseudo-inverse technique with Singular Value Decomposition (SVD) to verify the accuracy of the developed approach. Finally, sensitivity analysis of the joint stiffnesses connecting the substructures is performed to determine their effect on vibro-acoustic performance of the structure. It is shown that the regularization method plays a significant role on the outputs of the sensitivity analysis. All the studies presented in this thesis, demonstrate that the methodologies developed in this thesis accurately predict the end-to-end vibro-acoustic performance of the heavy duty truck cabin and also they can be utilized effectively to identify and redesign the components that degrade the performance.