Acoustic characterization of automotive mufflers. Part II: validation of the numerical models by means of experimental data

Increasing interest is being paid to noise pollution of internal combustion engines and as a result, recent international standards imposed more severe limitations to acoustic emissions on engine manufacturers. In particular, the noise coming from gas-dynamic interactions has an important influence in determining the final noise level of the engine; as a consequence, the muffler design is currently being considered as one of the most important research threads for engine companies. Within this context, the 1D approach to numerical simulations, which has been successfully applied by industrial designers to the fluid-dynamic design of the engine, is considered to be inaccurate in the evaluation of the acoustic behavior of the muffler for medium-high frequencies. On the other hand, an extension of the applicability of these codes in the medium-high frequencies would be desirable. The direct advantage would be the use of the same software for the simulation of both the fluid-dynamic and acoustic performance of the engine.On these bases, a commercial 1D numerical code was primarily analyzed in terms of accuracy, computational cost and modeling capability of mufflers from the acoustic point of view. As a second step, two non-conventional approaches were developed in order to improve the prediction capabilities of the 1D code and to widen its frequency range of validity, as well. The base scheme of these new approaches was to extend the application of the traditional 1D nonlinear equations not only in the axial direction but also in perpendicular directions within the cross-section of the muffler, achieving a simplified description of the acoustic phenomena in the whole volume.The 1D predictions using these new approaches were compared both with several sets of experimental data collected on a purposefully developed test rig and specific 3D simulations; as a result, their limits in terms of accuracy were highlighted. Moreover, the introduction of the new sub-models increased the accuracy of the simulations and the frequency range of validity, leading to notable results with respect to traditional formulations of the problem.