High order large eddy simulation for the analysis of tonal noise generation via aeroacoustic feedback effects at a side mirror

Abstract

In this work, the flow around a side mirror and the resulting tonal noise generation are investigated using highly accurate compressible large eddy simulations. Avoiding tonal noise, which can be perceived as disturbing whistling sound, is a crucial target in automotive aeroacoustics. However, the underlying mechanisms are not completely understood and can typically not be captured with state of the art computational aeroacoustics solvers used in industry. Acoustic feedback effects known from tonal airfoil self-noise are a possible cause at smooth mirror housings that exhibit laminar separation upstream of the trailing edge. Since this application demands high accuracy, a simulation code based on the high order discontinuous Galerkin spectral element method is employed. To enhance geometrical flexibility, it is augmented with an extension to non-conforming curved elements in three dimensions. In the first part of the investigation, the simulation framework is used to analyze an early development stage side mirror exhibiting tonal noise generation. Adopting the corresponding experimental configuration, the study considers an isolated side mirror mounted on the wind tunnel floor. The computational flow field is shown to agree remarkably well with the experimental one based on comparisons with static wall pressure, hotwire and PIV measurements. Discrete peaks are obtained in the computational acoustic spectrum, originating at the trailing edge of the mirror downstream of laminar separation. The identified tonal noise source regions match the experimental ones and quantitative agreement is achieved for one of the tonal peak frequencies. Perturbation simulations reveal global acoustic feedback instabilities selecting the same discrete frequencies observed in the developed flow. The feedback loop comprises convective disturbance growth in the separated shear layer, scattering at the trailing edge and reinforcement through receptivity to the emitted sound in the upstream boundary layer. In a second step, this mechanism is studied in more detail based on a specifically designed simplified two-dimensional model. A subdomain approach is introduced to exploit the two-dimensional shape and circumvent the computational cost associated with the bluff body wake of the model. Simulations of a range of free-stream velocities exhibit tonal frequencies varying similarly to the experimentally observed so-called 'ladder structure'. The tone frequencies are shown to evolve according to a theoretical feedback model based on linear stability theory. Finally, the efficacy of various modifications to the mirror contour to eliminate tonal noise generation is evaluated. The present work contributes to the understanding of tonal noise generation mechanisms and can guide future designs. Moreover, it corroborates the capacity of the present discontinuous Galerkin framework to accurately capture relevant but delicate aeroacoustic effects at complex geometries.