Modelling of kerf width and surface roughness using vibration signals in laser beam machining of stainless steel using design of experiments

Manufacturing industries are increasingly interested in laser beam machining as an efficient material removal process. Accordingly, there is great motivation in the modelling of the machining process to understand the physics behind the machining process in order to improve the machining efficiency. The present study made an attempt to predict surface roughness, kerf width and metal removal rate by utilizing the vibration signals measured during the laser beam machining of AISI 304 stainless steel. As per Taguchi's L9 design of experiments, nine experiments were conducted at three levels of laser power (2.0, 1.5 and 1.0 kW), laser frequency (10, 8 and 6 kHz), cutting speed (3.0, 2.5 and 2.0 m/min) and nozzle tip distance (1.2, 1 and 0.8 mm) on 6 mm thickness of sheet metal and the vibration of sheet metal was measured using an accelerometer. It was observed that, the sheet vibration along the cutting direction increased the metal removal rate and sheet vibration perpendicular to the cutting direction caused surface roughness. Mathematical models were developed with the vibration data and predicted the surface roughness, kerf width and metal removal rate. The root mean square error of the responses was calculated as 0.615 mu m, 5.44 mu m and 0.259 m3/min for surface roughness, kerf width and metal removal rate respectively, demonstrating a significant correlation with experimental results. Furthermore, effect of laser power, laser frequency, cutting speed and nozzle tip distance on the surface roughness, kerf width, heat affected zone and metal removal rate was studied. The laser power and cutting speed have significant effect on these responses. Finite element modelling based simulation was carried out and studied effect of molten metal temperature on the heat affected zone. Temperature in the molten pool reached to maximum of 1773 K and the heat affected zone increased, when the laser power increased to 2.0 kW. The root mean square error between the measured and simulated heat affected zone was computed as 4.68 mu m, indicating a good correlation between both.