Assessment of Flow Noise Mitigation Potential of a Complex Aftertreatment System through a Hybrid Computational Aeroacoustics Methodology

Flow noise produced by the turbulent motion of the exhaust gases is one of the main contributions to the noise generation for a heavy-duty vehicle. The exhaust system has therefore to be optimized since the early stages of the design to improve the engine's Noise Vibration Harshness (NVH) performance and to comply with legislation noise limits. In this context, the availability of reliable Computational Aero-Acoustics (CAA) methodologies is crucial to assess the noise mitigation potential of different exhaust system designs. In the present work, a characterization of the sound generation in a heavy-duty exhaust system was carried out evaluating the noise attenuation potential of a design modification, by means of a hybrid CAA methodology. In a first step, a steady state 3D-CFD simulation of the exhaust system in its baseline configuration was carried out with a RANS approach, to gather an analysis of the flow inside the diffusor and to obtain the turbulence intensity distribution necessary to localize and quantify the noise sources. Then, in a second step, the Stochastic Noise Generation and Radiation (SNGR) method was employed to synthetize the noise sources for the subsequent computation of the radiated acoustic field. A sensitivity analysis on the far field noise to the main method parameters was also performed, especially on the noise source region extension. Moreover, the baseline design of the exhaust system was also studied with a Direct Noise Calculation (DNC) approach, providing absolute flow noise levels to be compared with the results obtained by the means of the hybrid CAA approach. Then, a modified version of the exhaust diffusor was analysed with the proposed hybrid CAA methodology, highlighting the impressive potential in terms of noise attenuation of the new design configuration. The adoption of proposed hybrid CAA methodology was therefore demonstrated to allow a dramatic downscaling of the computational cost compared to DNC simulations, being fully compatible with the limited time available for the development of a new product in the automotive industry.