Tribological behaviour of ZA/ZrB2 in situ composites using response surface methodology and artificial neural network

Abstract

Abstract Recent advancement in metal matrix composites shows the reduction in resource and energy consumption through improvement in tribological properties. However, statistical modelling helps to achieve the material efficiency through optimizing experimental parameters. This study focuses on developing a statistical modelling to predict the tribological behaviour of ZA/ZrB2 in situ composites. Analysis of variance (ANOVA) was conducted using Response surface methodology (RSM) by Design expert 13 software which was suggested the use of quadratic model and the regression equation was developed. Varying load, sliding distance, and volume % of ZrB2 as input parameters and wear and coefficient of friction (COF) as response parameters were utilized in RSM. Results indicate that the volume % of ZrB2 is the main contributing parameter for wear while for COF, load is the main contributing factor. Artificial neural network (ANN) developed using PYTHON was also employed to validate the optimized parameters for wear and COF suggested by RSM. Experimental results and statistical analysis results obtained from RSM and ANN show the close tolerance and thus suggest a significant model that envisages the tribological characteristics of alloy and composites. Regression equation could be successfully utilized to predict the wear and COF at any given set of input variables based on applications.