Artificial neural network approach for casting process parameters on wear rate of Al-Cu-SiCp metal matrix composite

Abstract

In material selection, the prime parameter is the wear rate of material, mainly those engine and machine parts that are in dynamic condition. Stir casting is one of the cost-efficient and common routes for fabricating metal matrix composites (MMCs). The wear resistance properties of Al6061-Cu-SiC MMCs at variable levels of casting process parameters are investigated by the Pin-on-disc test. With the 5 levels of pouring temperatures ranging between 675 °C to 775 °C and 5 levels of stirring speeds in the range between 50 rpm and 800 rpm, experiments were conducted on three MMCs with the composition of Al-6061,4% Cu and 5% SiC particles with three varying particulate sizes as input parameters for experimental purpose. The sizes of reinforced SiC particulate in MMCs are 63 µm, 53 µm, and 45 µm respectively. The influence of material type on wear rate is found to be statistically significant using ANOVA. Although the temperature of pouring and the speed of stirring had a significant impact on the rate of wear of the material, it was found thatthe combined effect of the temperature of pouring and the material type, as well as the combined effect of the type of material and the speed of stirring, had no effect on the rate of wear. The rate of wear is not greatly affected by the combination of the temperature of pouring, material type, & the speed of stirring. Stirring speed and pouring temperature had a major effect on the wear rate (p = 0.00 ≫ 0.005).The worn specimen was analyzed by Scanning electron microscope to recognize the features of MMCs at independent variables and an Artificial Neural Network (ANN) were modeled to establish the relationship and prediction of wear rate at different input variables.