Grinding Temperature Prediction Modeling Framework Based on Single-to-Multi Grain Interaction Mechanism

Abstract Grinding has been an extremely fundamental finish/superfinish machining method to obtain high-dimensional accuracy and superior surface integrity for mechanical products. Generally, grinding temperature significantly influences the final product performance because it directly determines the thermal conditions and thus deformation and distortion of ground surface generated from grinding process. Grinding temperature has been mainly studied or predicted by means of theoretical analysis, experimental measurement, simulation model, and data-driven computer method. These model/methods have helped to reveal the basic mechanism, consider complex condition influence, and achieve better prediction accuracy. However, each individual method has certain drawbacks including low computational efficiency, inaccurate prediction, and high costs of data collection. It thus keeps as a great challenge to predict grinding temperature accurately and efficiently based on the micro-mechanics of grain-workpiece interaction. This research aims to develop a multi-scale ground surface temperature prediction model by accounting for the micro grain-workpiece contact behavior, the single grain temperature, and the motion state of multiple grains in the grinding zone. Firstly, the contact interaction between single grain and workpiece is equivalently transformed into a plane-cutting process between different micro-cutting edges of grain and ground surface. Then, a finite element model of single grain grinding is established to predict the grinding temperature distribution based on the geometry of one single grain. Meanwhile, the single grain temperature is calculated and accumulated to obtain the grinding temperature generated by reconstructing the practical grinding wheel morphology. The proposed model takes into account the random grain size, position distribution of grains, and grinding parameters to ensure its accuracy and credibility with sufficient predicted accuracy. The prediction error keeps within 10.4% by comparing the predicted and the measured grinding temperatures at the grinding depths from 10μm to 50μm. The novel approach reveals how the micro grain-workpiece interaction mechanism governs grinding temperature distribution and its evolution pattern.