Tool Wear in Cutting Operations: Experimental Analysis and Analytical Models

The possibility of predicting the amount of the tool wear in machining processes is an interesting topic for industries, since tool wear affects surface integrity of the final parts and tool life is strictly connected with substitution policy and production costs. The definition of models able to correctly forecast the tool wear development is an important topic in the research field. For this reason in the present work, a comparison between response surface methodology (RSM) and artificial neural networks (ANNs) fitting techniques in tool wear forecasting was performed. For developing these predictive models, experimental values of tool wear, obtained by longitudinal turning operations with variable cutting parameters, were collected. Once selected, the best configuration of the two previously mentioned techniques, the resultant errors with respect to experimental data were estimated and then compared. The results showed that the developed models are able to predict the amount of wear. The comparison demonstrated that ANNs give better approximation than RSM in the prediction of the amount of the flank wear (VB) and of the crater wear (KT) depth. The obtained results are interesting not only from a scientific point of view but also for industries. In fact, it should be possible to implement the best model into a production manager software in order to correctly define the tool change during the lot production.