Abstract
The robustness of brake systems with respect to squeal noise is a very important topic. It is well known and understood that variations on the surface of brake pads have a large influence on noise occurrence. Until now, however, automotive suppliers and OEMs consider these variations in a very limited way only. The reasons are straight-forward: The creation of a database of measurements is time-consuming and expensive. Second, the transfer of the variation shapes into a CAE model is a complex process requiring expert knowledge and a lot of manual work today. The objective of the collaboration of DAIMLER and DYNARDO is to fill this gap and establish a closed and easy-to-use workflow to translate measurement-into CAE-models for brake squeal robustness analysis. The geometric imperfections of the pad surfaces can be obtained using high-resolution laser scans. Their shapes are very different among the individual brake pad specimen depending on the pad model, used pad materials, driving conditions and environmental factors. A detailed and accurate description of the variation in geometric scatter is important for accurately predicting contact areas, contact pressures and, finally, noise phenomena. Nevertheless, by using a decomposition into scatter shapes and amplitudes it is possible to create a statistically equivalent model for the representation of geometric imperfections once a sufficiently large number of measured specimen is available. Such random field models can be created up to any scale of spatial resolution allowing a great accuracy in representing the geometric shape patterns as found in the measurements.
This article presents a new automated workflow which supports engineers in transferring the measurements to their CAE/FEM based robustness analysis. Once established, the workflow can be used to either automatically morph the FEM meshes to match the measured geometries 1-to-1. Second it can be used to generate new virtual pad geometries which obey the spatial variation patterns and probability distributions obtained from the whole set of measurements. Since in practice the number of measurements is quite small for the same pad geometry, the database allows to import and transform the variation shapes of different pad geometries to a single reference. Further, it also models the correlations between the variations of the outer and the inner brake pads. Typically, the variation shapes of the inner and outer side are very different from each other. Finally, the FEM model of the brake system may be changed regarding the geometric parameters of the considered pad, the used FEM software and the FEM meshing parameters. All these issues must be considered in the design of the workflow. In this study, the workflow is applied to ABAQUS, but can be used in conjunction with other solvers (e.g. ANSYS, LS-DYNA, NASTRAN). DYNARDO develops the software Statistics on Structures (SoS). It is used to analyse the measurements and morph the FEM meshes. The software optiSLang is used to organize and automatize the workflow and to perform the robustness evaluation. The FEM modelling and measurements are implemented by DAIMLER. They are used to demonstrate the influence of the imperfections onto the brake squeal noise and to validate the simulation results with respect to the measurements.
KEYWORDS Robustness analysis, Geometric uncertainties, Squeal noise, Laser scans, Random fields