Predicting wear rates in magnesium-based composites reinforced with Ti and SiC particles using artificial neural networks
Hossein Ahmadian, Sabbah Ataya, Waleed Mohammed Abdelfattah, A. Fathy
Abstract
This study presents experimental investigations and models for prediction of wear rates of magnesium (Mg) composites reinforced with titanium (Ti) and silicon carbide (SiC). Mg exhibits low density and high thermal conductivity, making it suitable for weight-sensitive applications; however, its inherent wear resistance is limited. By adding Ti and SiC, both the mechanical and tribological properties improved. The experiments employed high energy milling process at different milling times from 2 to 16 h and varied the Ti content between 15 and 30 wt% and SiC between 5 and 15 wt%. These factors affected hardness (108–137 HV), density (1.65–2.15 g/cm 3 ), and porosity (16.31–30.58 %). It was found that long milling times made the material more porous and softer, which reduced wear resistance. The presence of Ti in the microstructure reduces the porosity, while SiC increased hardness. Wear tests at loads of 2 N, 4 N, and 8 N gave wear rates between 0.01 and 0.3 mm 3 N −1 m −1 . A genetic algorithm (GA) was also used to improve an artificial neural network (ANN) for predicting wear rate. The improved ANN-GA model reached a least-square value of 0.9629, compared with 0.9234 for the basic ANN. In general, the results show that optimizing milling time and reinforcement balance is important for improving the wear resistance of Mg–Ti–SiC composites.