Multi-response ANN modelling and analysis on sliding wear behavior of Al7075/B4C/fly ash hybrid nanocomposites

Abstract

The accumulation and disposal of fly ash is a common problem in thermal power plants leading to environmental concern. In this work, boron carbide (B4C) along with fly ash is added as reinforcement particles in the aluminium matrix to fabricate a new class of composite with reduced density and improved wear resistance. The hybrid metal matrix nanocomposites (MMNC) having Al7075 matrix alloy reinforced with B4C (1.5 wt%) and fly ash (0.5, 1.0 and 1.5 wt%) nanoparticles were produced by ultrasonic stir casting method. The microstructure study of the composite showed uniform dispersion of particles in the matrix and absence of unwanted carbides. The dry-sliding wear test was performed by varying weight % of fly ash (R), load (L), sliding distance (D) and sliding velocity (V) as input factors; to study wear rate (WR) and coefficient of friction (COF) being the output parameters. The experimental result is analyzed by response surface methodology (RSM) while the wear behaviour predictive model is developed using artificial neural networking (ANN). The ANN model with 4-6-2 network architecture is found effective with an average percentage error of 9.32% and 3.80% for WR and COF respectively. Analysis of variance (ANOVA) result established applied load as the significant parameter affecting both the responses (76.21% for WR and 79.55% for COF) followed by % fly ash (12.77% for WR and 9.62% for COF). A good enhancement in the wear properties of all combinations of reinforcements in Al7075 hybrid nanocomposites is observed. The hybrid nanocomposites having 1.5 wt% of B4C and 1.5 wt% of fly ash exhibited the maximum wear resistance.