A methodology to study automotive brake and dry clutch NVH

The automotive industry faces a continuous challenge to design vehicles that meet increasingly stringent regulations along with societal expectations. Often, any advancement towards meeting the prescribed design objectives may, without a fundamental understanding of the system-level behaviour, worsen the performance of the vehicle in another respect, creating Noise, Vibration and Harshness (NVH) issues. The development of physics-based transient dynamic models, of friction disc powertrain subcomponents (clutches and brakes), capable of predicting errant NVH behaviour is a highly desirable tool for the automotive industry to meet such contradictory design requirements.

The aim of the current work is to investigate friction disc oscillatory response throughout the engagement process, link the frictional behaviour with the system dynamics and optimize the design parameters, using AI methods, for improving clutch NVH performance and oscillatory behaviour. A transient clutch dynamics model, which can accurately simulate typical engagement manoeuvres, is developed to identify potential unstable behaviour due to component unwanted motions. The coupling between torsional, tilting, bending and radial degrees of freedom, of the clutch components, is accomplished through the frictional interactions between component interfaces (clutch disc and pressure plate). The transient model is validated against experimentally obtained data from a vehicle and it utilizes coefficient of friction data, measured under representative interfacial conditions (pressure and relative speed) using a pin-on-disc machine. The model predicts clutch dynamic instability behaviour including: mode coupling instabilities along with the related system frequencies and mode shapes. A parametric study shows that the variation of geometrical parameters (pressure plate mass/clutch disc mass as well as the variation of system characteristics (stiffnesses and coefficient of friction) can lead to more than one mode coupling instability during the engagement process. The results from the parametric studies were used to develop a design optimisation process of the key system parameters (utilising a Recurrent Neural Network) for improving the NVH performance.