Optimization and Prediction of Surface Roughness Profiles of Machined Heat affected Zone of Mild Steel Weld Using Response Surface Methodology and Genetic Algorithms

Abstract

The service life of a weld fabricated engineering product is dependent on the surface finish of the product. Research has revealed that most of the failures observed in fabricated metal structures is linked to excessive heat input and large heat affect zone. This study is applying response surface methodology and genetic algorithms to optimize and predict the surface roughness of machined heat affected zone of mild steel welds. The design expert software was employed to produce a design matrix using the range and level of the input parameters. The central composite design (CCD) was used. 30 sets of experiment are performed according to the design of experiment; the input parameters are cutting speed, feed rate, nose roughness and chip thickness. 2 analytical methods are employed namely RSM and GA. From the results obtained, the ANOVA showed that the second order polynomials are suggested as the best fit to predict the large response, contour plot and surface plot showed the interaction between the cutting speed, feed rate and the surface roughness. The metals developed have high strength and adequately. Results obtained in this study showed that the interactive combination of nose radius and depth of cut has a very significant influence on surface roughness and chip thickness. The variance inflation factor has a value of 1 for the independent and combined level of the input factors. The model had a coefficient of determination value of 93% for surface roughness.