The importance of braking system in vehicle safety is undeniable. Due to the safety requirements and also the social and political regulations, the braking system should fulfil many design criteria, e.g. long-lasting brake pads and silent braking. In various braking situations like prolonged low speed braking, as it often happens during parking manoeuvres, a disturbing squealing noise occurs frequently.
The brake squeal has an adverse impact on the perception of safety and quality. Hence, among all NVH related phenomena in ground transport vehicles, understanding and modelling the squeal noise plays a significant role in the design and optimization of braking systems. The presented work attempts to model the squeal noise using a semi-analytical approach.
In many available experimental methods, the squeal noise will be measured at a pre-defined set of input parameters, coming from a dyno test rig. These parameters are mainly braking pressure and temperature, initial velocity and braking momentum. The output of these measurements, however, covers a broad range of sound levels and frequencies. To analyse the output parameters a new algorithm was developed and implemented into the BSD-Tool. The algorithm distinguishes real squeal noise from background noise of each measurement. These filtered data then in combination with the use of the presented semi-analytical model will generate more promising analysis and results, by ignoring measured background noise. Though, the measured frequencies seem to be hardly reproducible, introducing new hidden parameters can facilitate correlating the measured noises with all the input parameters.
In the presented work, large experimental data sets have been obtained for squeal noise. In order to prevent any superfluous and undesirable algebraic fitting, however, only a small part of these data have been analysed. For this data set, important dimensionless numbers as characteristic noise and resonance amplitude, brake pad Péclet number, brake pressure friction factor, brake thermal diffusivity, and characteristic roughness amplitude have been introduced. Applying these dimensionless numbers, a rough correlation for noise has been attained. This correlation was able to detect the trend of data and interestingly was able to cluster them roughly.
Though the correlation shows the significance of some operational parameters such as the dynamic friction factor on the brake noise, it shows inherently the necessity of introducing hidden parameters which may describe the complex behaviour of noise in a rational means.
Considering the stick-slip effect on noise development, a modified static friction factor as the preliminary hidden parameter has been defined. Besides the enhancement of noise prediction, this modified parameter was able to describe further phenomena, i.e. stick-slip-hysteresis effect and elasto-plastic deformation. These effects seem to be very essential for brake noise. In order to include those, two new hidden parameters have been specified. Due to the lack of experimental methods, these parameters have been developed using analytical approaches.
Applying all abovementioned conventional and hidden dimensionless parameters, the final semi-analytical model of the presented work was able to correlate, recognise and cluster all the available data. This correlation can be used for brake noise analysis and also for development of new experimental methods, as well as the enhancement in brake pad design processes.
Mohammadi Koorosh, Hering Kurt, Goecke David, MPlan GmbH, Weissach, Germany