Development of a 3D-CFD Methodology for the Aeroacoustic Analysis of a High-Performance SI Engine Exhaust Line

Sound plays an important role in the automotive sector, especially for sports cars where it is used as a signature of manufacturer's identity and exclusiveness. In order to enhance engine sound quality and to meet increasingly stringent noise emissions regulations, robust development methodologies are needed since early engine design phases. 3D-CFD analyses, in particular, could provide significative information concerning sound generation and engine signature. The present thesis work, conducted at POWERTECH s.r.l in collaboration with Maserati S.p.a., focuses on the development of a 3D-CFD aeroacoustic methodology for the assessment of sound generation mechanisms in a high-performance SI engine exhaust line using the CFD software StarCCM+. The requirements for aeroacoustic simulations differ from standard aerodynamic calculations, making them much more challenging. For this reason, a specific setup is required for a correct resolution of the turbulent flow field and for the description of the acoustic propagation. The Direct Noise Calculation approach has been adopted for the description of flow development, sound generation and propagation in the near-field, while the Ffowcs Williams and Hawkings (FWH) analogy has been used to account for sound propagation in the far-field. The closure of the FWH surface has been studied showing differences in terms of spectrum and directivity between closed and open options. The exhaust gas was firstly treated as a multi-component fluid, the subsequent simplification as a single-component gas showed significant reduction in the computational cost with negligible effects on fluid dynamics and acoustic field.\\In order to validate the methodology, sound pressure level of the principal engine orders obtained in an experimental test have been compared with those calculated in the 3D-CFD simulation. The comparison shows the good agreement of CFD results with experimental data.