| Noise level is one of the important performance parameters of automobiles, which affects the driving comfort. With the development of the expressway, the increasing of the automobile speed, automobile aerodynamic noise at high speed has been a more and more important component in the whole automobile noise sources by using advances in reducing the engine noise, tire noise and other noises. It has been becoming a state-of-the-art, prominent topic to control the automobile aerodynamic noise.The new automobile style development cycle has been becoming shorter and shorter. Correspondingly, there was a trend to carry out aerodynamic noise study at the early stage of the development cycle. Many automobile manufactures reduced the leak noise greatly by improving sealing performances, however, aerodynamic noise generated by unsteady airflows outside a vehicle still remains as problems to be solved. Because the aerodynamic noise is generated by the transient features of the flow field around the vehicle, it is necessary to study aerodynamic noise characteristics and generation mechanisms based on the understanding of flow characteristics about each part of a vehicle.Main methods in researching aerodynamic noise are experiments and numerical simulations. Wind tunnel experiments are the dominant method among them. And the on-road measurements could supplement the wind tunnel experiments. Engine noise, powertrain noise and tire noise are not present when testing in a wind tunnel. Also, the wind tunnel has the advantage that the flow speed and direction can be held constant. Hence, wind noise measurements made in a wind tunnel are very repeatable, and small increments can be discerned in the measured wind noise level. However, aeroacousitc wind tunnels are really expensive to build and expensive to rent time in, a great deal of engineering demands could not be fulfilled only by wind tunnel experiments. Numerical simulation methods have been applied in predicting aerodynamic noise with the rapid development of the Computational Fluid Dynamics (CFD) and computer hardware. The critical factor in noise numerical simulation is how to obtain accurately transient information of the exterior flow field, especially the transient separated flow characteristics caused by local shape details of an automobile. Most of research objects are simplified shapes because the computational load using traditional numerical method is really huge. And the large scale vortex structures in the wake of blunt bodies could be described. Generally, these wake structures have little or no effect on aerodynamic noise although they influenced the aerodynamic drag greatly. Therefore, transient flow features of a real vehicle could not be obtained by using simplified shapes. And aerodynamic noise information could not be obtained correctly.Based on previous research work, the features of recent numerical simulation methods were reviewed and summarized. The subdomain numerical simulation method is brought up focusing on the difficulties in predicting the aerodynamic noise. Firstly,the steady Reynolds Averaged Navier-Stokes (RANS) method is applied in the solution of the whole automobile exterior flow field. Then, the critical region was extracted from the whole domain as the subdomain, and the mesh of the subdomain was refined. The steady solution results were mapped as boundary conditions and initial field conditions to the subdomain. Finally, the subdomain flow field and aerodynamic noise were solved using Large Eddy Simulation (LES) method. This method reduced the computational load effectively without accuracy compromise.The exterior flow field around an aerodynamic reference notchback car model was simulated using steady RANS method. This method was proved valid according to the comparison between the solution results and wind tunnel experimental data in previous studies. Then two groups of rear view mirrors were designed and assembled with the car model in order to study the influence of the geometrical shape parameters on the aerodynamic drag and aerodynamic noise. It was illustrated by computational results that the geometrical shape parameters of the rear view mirror have little effect on aerodynamic drag, which made good agreement with related research conclusions. Based on the RANS results, the aerodanamic noise source items were calculated using Lilley Equation and Synthetic Turbulence Method. And the boundary layer noise power level was predicted using Lighthill-Curle Equation. Influences of the angleδ(between the rear view mirror profile and the car model) on the aerodynamic noise were analysed. The noise source items when angleδranging from -10°to 0°were stronger than the items from 0°to 10°.The region of A-pillar and rear view mirror is a critical area as far as automobile exterior aerodynamic noise is concerned. A simplified automobile greenhouse model was designed for simulating the flow status around A-pillar. This model was also the base for mounting the rear view mirror and for validating the subdomain numerical simulation method. Two rear view mirrors were mounted on the simplified greenhouse model respectively. And the whole domains including the greenhouse model and mirrors were solved using the RANS method which was proved in previous study mentioned above. The rear view mirror wake was selected to be the subdomain according to the RANS results. Then the transient flow field in the subdomain was computed using LES method. It was illustrated by the results that the main reason of aerodynamic noise in the A-pillar and rear view mirror region is the transient wake of the mirror. And the frequency feature is broadband. The aerodynamic noise level generated by the new rear view mirror was lower than the base one based on comparing nine points results in the mirror wake. The same phenomena are also predicted by steady RANS results.At present, another important exterior aerodynamic noise region is roof rack crossbars on the SUV and crossovers. The aerodynamic noise generated by the rack crossbars was simulated. And the influences caused by different cross section shapes and mounting angles were also analysed. It was shown by the computational results that the narrow band aerodynamic noise, which was generated by transient two dimensional separated flows, could be reduced by the simple streamlined cross section providing the attached flow on the roof. Conclusions in related studies were revised by numerical simulation results. The aerodynamic noise would not be severe when the angle of attack ranged from -8°to 4°. The three dimensional flow around the rack was simulated by the subdomain method based on the two dimensional solution. And the influences of different mounting styles of the crossbar ends on aerodynamic noise were analysed. It was illustrated that the crossbar which was extended outside of the rack rail would cause trailing vortex. And it would generate high level and broadband aerodynamic noise.The experimental validation process of the aerodynamic noise numerical simulation consists of two steps. Firstly, transient flow characteristics predicted by numerical simulation should be validated by experiments. Secondly, the accuracy of aerodynamic noise computation would be analysed. The automobile aerodynamic noise was studied using the first low background noise automotive wind tunnel in China. The qualitative experiments including surface tufts method and oil flow method were carried out for visualizing the flow field. And the numerical simulation methods were validated for describing the transient characteristics of the flow field by comparing with the wind tunnel experiments. The transient velocity fields of the rear view mirror wake and the rack crossbar wake were measured using Particle Image Velocimetry (PIV) Technique. And it was illustrated by the experimental data that the vortex scale was comparatively large in the wake of rear view mirror and typical blunt body's wake structures were represented in the time averaged velocity field. On the contrary, the vortex scale was comparatively small in the wake of rack crossbar and typical Von Kármán Vortex structures were recorded in transient velocity field images. Numerical simulation methods were proved to be correct by comparing the quantitative experimental data and computational results.Aerodynamic noise generated by rear view mirror was measured using acoustic sensors which were mounted on the surface of the greenhouse model. It was shown that the wind tunnel experimental data made good agreement with numerical simulaition results at the mid-frequency range (500Hz~2500Hz). And there were obvious differences between the experiment and the numerical simulation at the low frequency range (lower than 500Hz). The main reason of the differences was the background noise of the wind tunnel was also at low frequency range.Aerodynamic noise caused by rack crossbar was measured in the wind tunnel. It was not easy to fix sensors because the monitor points were not on the surface of the model. The nose cones were added in order to reduce the self-induced noise by the sensor. The trend of the sound pressure level (SPL) frequency signal obtained by experiments and numerical simulations was the same. However, there were some differences at the frequency data which were corresponding to the narrow band SPL peaks. The main reasons of the differences were the unstable features of the Von Kármán Vortex generated by rack crossbar and the self-induced noise of the sensor could not be eliminated completely. Thus, the SPL peak frequency might change in certain range.The subdomain numerical simulation method for predicting automobile exterior aerodynamic noise was proved to be reliable by wind tunnel experiments. Aerodynamic noise characteristics of a light-duty van and a SUV were studied using the subdomain method. It was shown that the aerodynamic noise generated by rear view mirror of the light-duty van could be reduced by changing the mounting angle in the legal function range of the mirror. It was represented that the aerodynamic noise generated by crossbars with simple streamlined cross section was much lower than the noise generated by traditional cross section. The subdomain numerical simulation method could be applied for evaluating aerodynamic noise and providing recommendations for shape design to reduce noise at the early stage of the new product development.In summary, the subdomain numerical simulation method for predicting automobile exterior aerodynamic noise was put forward based on systematic research. Main aerodynamic noise sources positions were identitied and the SPL signals were obtained at monitoring points using this method. The generation reason and mechanism of aerodynamic noise were revealed by numerical simulation results and wind tunnel experimental data. Evidences were provided for reducing aerodynamic noise at the early stage of new product development. And research conclusions were conducive to the engineering cases.
|
PDF Full Text Request