A Statistical Model of Laser Surface Finishing Using Design of Experiments and ANOVA

In previous work, laser polishing indirect-SLS metal parts was achieved using a Nd:YAG laser raster scanned at high speed. This work showed surface roughness could effectively be reduced by means of shallow surface melting. Surface area roughness data, Ra, was obtained by line profilometry measurements of the polished samples according to a modified procedure based on the ASME standard for surface texture measurements. A Design of Experiments (DOE) was conducted to better understand the effects of processing parameters such as laser beam power, scan time, and scan line density on the surface roughness. The DOE consisted of a multi-level test for each variable and included one replicate. ANOVA was used to determine the significance of each variable and the interactions between variables. Three reduced order models were derived, but large variations in absorptivity in the experiment resulted in large errors and inaccurate models. The data did show the average reduction in roughness over all the trails was about 20%, however several trials resulted in roughness reductions of over 50%. Future work will concentrate on reducing the absorptivity noise to produce more accurate models.