Research article on thermal evaluation of Al-Si3N4 composites using Machine Learning

Abstract

Lightweight aluminium alloys are being rigorously researched for their growing industrial applications because of their significant property advantages and cost-effectiveness. This work focuses on producing novel composites based on the Al6061 alloy by adding reinforcement at different weight concentrations of 0, 3, 6, 9, and 12 % (silicon nitride) via liquid metallurgy stir casting. To determine the impact of reinforcing percentage on Al6061 alloy, the microstructure, EDS analysis, thermal conductivity and thermal expansion of these stir cast specimens were evaluated. Thermal conductivity was evaluated using a comparable cut bar method (ASTM E1225), whereas thermal expansion was determined using a push-rod dilatometer method (ASTM E228). From experimentation, it was observed that microstructure analysis of composite showed uniform distribution of Si3N4 micron particles with the smallest of porosities and agglomeration was observed. The formation of dendrite structure was observed in all composite specimens due to faster solidification. The thermal conductivity increased by 23.33 %, and the coefficient of thermal expansion decreased by 20.73 %. The thermal characteristics of Al6061 reinforced with Si3N4 micron particles were superior to those of Al6061 without Si3N4 reinforcement. Machine Learning models were used to predict the coefficient of thermal expansion and the thermal conductivity of Al 6061 composites. Different models were trained with the experimental datasets and the performance of each model was evaluated.