Numerical and statistical modelling of high speed rotating diode laser surface hardening process on a steel rod

Abstract

The present study aims to develop an integrated statistical mathematical model utilizing response surface methodology (RSM) coupled with finite element method (FEM) for optimizing laser surface hardening process on a cylindrical steel solid rod. The proposed quasi-stationary laser beam processing technique utilizing high-speed rotation facilitates to obtain a uniform hardened layer along the entire work piece surface with complete elimination of deleterious inter-pass tempering effects in overlapped regions. The developed model validated with experimental results conducted by changing laser power (2500–3500 W), linear speed (8–12 mm/s) and rotary axis speed (1600–2000 RPM) on a 12 mm thick En-31 steel cylindrical rod employing a 4 mm X 4 mm square laser spot. The setup constitutes a 1.0 mm fiber-coupled diode laser integrated to 6-axis Robot and a high-speed rotating lathe. The three-factor three-level Box-Behnken design available in RSM used to develop the regression model. A multi-objective optimization technique used to optimize the process parameters on attaining maximum case depth, maximum mean hardness along hardened layer depth and minimum differential hardness across the treated layer. Good agreement realized between actual experimental values to optimized values obtained from the RSM model with desirability approach resulting in complete elimination of tempering effects in overlapped regions. Indeed, the optimal processing parameter combination obtained from RSM methodology desirability approach constituted 3500 W laser power, 8 mm/s linear speed and 2000 RPM that facilitated in validation of hardened layer in terms of 330–350 µm case depth with minimal differential hardness of 56–58 HV within the treated layer throughout its length processed. Apparently, optimum parameters obtained with a desirability approach found to be in good agreement with thermal profiles obtained from FEM analysis with an error difference of 3–5%.