Predicted Roughness Perception for Simulated Vehicle Interior Noise 2012-01-1561

In the past the exterior and interior noise level of vehicles has been largely reduced to follow stricter legislation and due to the demand of the customers. As a consequence, the noise quality and no longer the noise level inside the vehicle plays a crucial role. For an economic development of new powertrains it is important to assess noise quality already in early development stages by the use of simulation. Recent progress in NVH simulation methods of powertrain and vehicle in time and frequency domain provides the basis to pre-calculated sound pressure signals at arbitrary positions in the car interior. Advanced simulation tools for elastic multi-body simulation and novel strategies to measure acoustical transfer paths are combined to achieve this goal.

In order to evaluate the obtained sound impression a roughness prediction model has been developed. The proposed roughness model is a continuation of the model published by Hoeldrich and Pflueger. Within the model simultaneous as well as temporal masking effects are considered. In addition, specific model parameters have been adjusted to predict subjective ratings of 18 experienced subjects, including mechanical engineers and audio engineers. The adapted roughness model has been developed by the usage of real sound stimuli measured in the car interior for different pre-defined engine types. Regression analysis shows that in most cases the subjectively perceived roughness can be predicted with good accuracy. Finally, the development model is tested with new stimuli not used in the development of the model; also for these new stimuli a good agreement of R₂ ≥ 88% could be achieved.

After the discussion of the roughness prediction model, parameter variations for an automotive internal combustion engine (ICE) are discussed and compared with the aid of the new roughness model. From the results it is shown that the developed model is well suited to assess design changes and their consequences on the perceived roughness. Therefore, it can be used to develop roughness optimized solutions already in early design stages.