Exhaust systems are a necessary solution to reduce combustion engine noise originating from flow fluctuations released at each firing cycle. However, exhaust systems also generate a back pressure detrimental for the engine efficiency. This back pressure must be controlled to guarantee optimal operating conditions for the engine. To satisfy both optimal operating conditions and optimal noise levels, the internal design of exhaust systems has become complex, often leading to the emergence of undesired noise generated by turbulent flow circulating inside a muffler. Associated details needed for the manufacturing process, such as brackets for the connection between parts, can interact with the flow, generating additional flow noise or whistles.
To minimize the risks of undesirable noise, multiple exhaust designs must be assessed early to prevent the late detection of issues, when design and manufacturing process are frozen. However, designing via an experimental approach is challenging. Since the construction process does not exist yet, physical prototypes lack the details associated to manufacturing. In addition, experimental optimization is time-consuming, as each design iteration will require a new physical prototype, thus increasing costs and development times, or limiting the explored design space.
Alternatively, larger design spaces may be explored using virtual optimization, while removing the limitations of physical testing. Exhaust flow and acoustic simulations with the Lattice-Boltzmann Method (LBM) have been shown accurate as well as feasible within the product design cycle timing. Using noise sources detection techniques, such as Flow-Induced Noise Detection (FIND), understanding the noise generation mechanism associated to the optimal designs is also possible to orient future design decisions. In this study, after validating the ability of the approach at capturing flow noise and whistles, a characterization of the bracket design connecting tailpipe and muffler is performed to minimize the risk of whistle in the final product.