Search

Zinc oxide (ZnO)-based varistor ceramic is a hard-to-machine material. Its microstructure, developed for its electrical properties, results in extremely low fracture toughness. Therefore, the current state-of-the-art in machining this material is lapping. In the present work, laser-assisted milling (LAMill) of ZnO ceramics was evaluated with a systematic approach to allow for new insights and advance the understanding of damage formation during machining of this material. A machining strategy to machine ceramic workpieces along the edge was studied. Additionally, a new transient analytical thermal model was developed to accurately predict laser-induced temperatures near the edge of the workpiece without excessive use of computational power, and the model predictions were experimentally verified. A novel approach to evidence machining-induced damage with the mathematical description of the ideal machined surface was also designed and tested. It was found that at higher temperatures the propagation of intergranular fractures is inhibited. Another mechanism based on the material porosity and contributing to more instances of intergranular crack initiation has been identified. To reliably predict the extent of damage on the machined surface, the analysis has been extended by covering the effects of the two identified influential process parameters: feed per tooth, f$_z$, and the temperature of the to-be-cut material, T. By adjusting these parameters, surface damage was successfully controlled and the extent of damage was minimized at f$_z$ = 0.18 mm and T = 380 °C, achieving a 44 % reduction compared to the conventional process with the same machining parameters.

Predictive models are crucially relevant for the design and control of manufacturing processes, particularly when process quality is highly sensitive to unknown physical parameters. This paper describes a method to model industrial processes using neural networks and physical prior knowledge. The approach is applied to heat transfer problems, particularly relevant for additive manufacturing processes, and the results are compared to alternative methodologies. Obtained results show that the integration of partial differential equations in the neural network model leads to reduced amounts of required training data and increases the model stability. Such outcomes represent promising characteristics for novel model predictive control strategies., Procedia CIRP, 102, ISSN:2212-8271, 18th CIRP Conference on Modeling of Machining Operations (CMMO)