Abstract
KEYWORDS –Brake noise, parameter identification, full brake system, instability prediction
ABSTRACT
Brake squeal is a strong cost driver for automotive manufacturers owing to warranty-claim related costs. It is difficult to model numerically because of nonlinearities, and uncertainties in complex contact interactions, friction characteristics, operating and environmental conditions in a brake system. In addition, finite element brake models are often not updated with experimental modal tests so that instabilities can be over-predicted or under-predicted. The objective of this study is to improve instability predictions using the complex eigenvalue analysis by applying model updating to a finite element brake model and to highlight the influence of each updating step on instability predictions. In this study, a finite element brake model is updated with experimental modal tests at the component level and the subassembly level. The effect of the abutment clip stiffness and the pad assembly on complex eigenvalue analysis is investigated. The pad is updated using a manufactured surrogate model of a square lining and a small backplate, which is correlated to a full pad with realistic geometry. Updated material parameters and an optimised Rayleigh damping model are used. A rigorous mesh refinement study and two different friction laws are also considered to explore differences in instability predictions using experimentally determined contact surface roughness profiles. Instability predictions using the complex eigenvalue analysis are compared with dynamometer squeal tests. Implications for further refinement of the updating approach are discussed. Results show that by applying a refined mesh to model more realistic contact and friction models, and updated material properties and damping, the prediction of instabilities is improved. It is shown that the updating of the abutment clip stiffness and the correct pad assembly including the shim has a significant influence on the instability predictions. Comparisons with experimentally determined contact surface roughness show that accurate contact area modelling between the pad with rotor is essential in order to improve the numerical model. As an outstanding issue, only the forced response of the full and pressurized system and its correlation with numerical results would provide a reliably updated model. While the successively updated model gives important insights into the importance of the individual steps, a rigorous sensitivity study as well as a forced response analysis would be an extension of the present research to fully validate the finite element model for improved instability predictions. While model updating itself is not new, this study highlights the importance of proper modelling of individual components including numerical aspects, contact and complexity such as validating subassemblies for the first time via comparing squeal predictions and brake noise measurements.