Abstract:This study presents a comprehensive Triboinformaticmodelling approach to predict wear and erosion behavior of Additively Manufactured (AM) AlSi10Mg alloy using Laser Powder Bed Fusion (LPBF) technology. The investigation encompasses three distinct build orientations: horizontal, vertical, and inclined, which were analysed using Machine Learning techniques, specifically K-Nearest Neighbors (K-NN) and Artificial Neural Networks (ANN). A dataset with a wide variety of microstructures and mechanical characteristics was produced by methodically varying the LPBF AM process settings. The AM AlSi10Mg specimens underwent surface modification before experimental wear and Heat-Treatment before erosion testing for each construction orientation. Predictive models were trained and validated using the resultant wear and erosion data, microstructural characteristics, and process parameters. K-NN was used because it is straightforward and simple to use, whereas ANN was used because it can capture complicated non-linear interactions. With amazing prediction accuracy, the models were able to identify wear and erosion patterns in various construction orientations. The ANN model offered improved accuracy for complex situations, whereas the K-NN model showed strong prediction skills with quick inference times. The potential of machine learning approaches in forecasting wear and erosion behaviour of AM AlSi10Mg alloy across different construction orientations is highlighted by this Triboinformatic study overall. The models used in this study help to improve our understanding of the tribo-mechanical characteristics of additive manufacturing (AM) components and make it easier to design and use AM-produced parts in settings that are sensitive to wear and erosion.

Anand Deshpande

Affiliation

Conference Tracks

Refine Results

Triboinformatic modelling of Additively Manufactured AlSi10Mg using Machine Learning