Process optimization for enhanced tribological properties of Al/MWCNT composites produced by powder metallurgy using artificial neural networks

Abstract In order to eliminate the agglomeration problem of reinforcement in the nanocomposite, a two-step dispersion process was employed. Under ultra-sonication and ball milling, 1 wt% of multi-walled carbon nanotubes (MWCNTs) were properly dispersed in pure aluminum (Al) (used as the matrix phase). The composite powder mixture was then consolidated in an inert Ar gas atmosphere by hot pressing under certain fabrication parameters. The powder mixture was characterized by Raman Spectroscopy, and it was found that MWCNTs did not cause structural defects in the pre-production process. The microstructural analysis of the sintered composites by Scanning Electron Microscope (SEM) and Energy-Dispersive x-ray Spectroscopy (EDS), revealed that the reinforcement was uniformly distributed in the matrix. Wear test results indicated that the wear resistance of the composites increased with increase of MWCNT reinforcement, and the wear mechanism was determined to be a mixing type by examining the wear traces by SEM. In order to determine the effects of different process parameters on wear loss, a Multilayer Perceptron (MLP) based Artificial Neural Network (ANN) was used, and experimental and predicted values were compared. It was noticed that the MLP based ANN model effectively evaluated the wear properties of the Al/MWCNT composites.