A novel grey-box based friction model for a wide range of machining conditions

Modelling the friction behaviour of cutting tools is a vital step towards understanding the complex tribo-mechanical system in cutting necessary for further improving coatings. However, measuring the friction behaviour during actual cutting is challenging due to its dependence on locally changing process conditions along the cutting tool such as sliding velocity and normal pressure. Thus this study introduces a novel tribometer to identify friction coefficients under a wide variety of normal pressures (914.7 MPa to 2170 MPa) and sliding velocities (20 m/min to 250 m/min) relevant for machining. Subsequently, the adhesive friction coefficient is determined inversely by modelling the experiments via Finite Element Analysis. The wear behaviour of coated pins is discussed for a wide range of contact pressures and sliding velocities relevant for cutting. A custom Python interface is presented which enables the local prediction of velocity and normal pressure dependent friction coefficients along the cutting edge within machining simulations. Common machine learning libraries can therefore directly be introduced in the FEA engine. Supervised machine learning regression models are trained and evaluated regarding their predictive capability. The Grey-Box model allows the AI-based local prediction of friction coefficients in cutting simulations based on the process conditions at the tool-chip interface. • Development of a novel tribometer for high sliding velocities and contact pressures relevant for metal cutting applications • Wear analysis of coated pins under varying process variables • Intelligent friction modelling with machine learning algorithms • Development of a python framework to predict friction coefficients along the cutting edge depending on local state variables