Surface characteristics investigation of 3D-printed PET-G plates during CO2 laser cutting

Abstract

This study investigates the influence of the stand-off distance (SoD), laser speed (LS) and laser power (LP) on kerf geometry (upper width Wu, down width Wd and kerf angle KA) and mean surface roughness (Ra) during CO2 laser cutting of 3D-Printed Polyethylene-Terephthalate-Glycol (PET-G) thin plates. A full factorial design is implemented, having two levels for SoD and three levels for LS and four levels for LP, resulting in twenty-four (24) cuts. The specimens were manufactured with the Fused Filament Fabrication (FFF) process and had a thickness of 4 mm. First, the Wu, Wd, and KA parameters were measured. Then, the specimens were cut into 24 smaller parts, and the Ra of the processed surfaces was measured. After analyzing the results using descriptive statistical analysis and contour plots corresponding to the two objectives of KA and Ra, a feed-forward and backpropagation neural network (FFBP-NN) was applied. After a series of subsequent training, it was found that the 3-10-2 NN architecture was quite efficient in explaining the variation in terms of the responses, thus having merit in practical applications related to laser cutting. The LC surfaces showed much better Ra (7.2-12 μm) than those in the literature for FFF parts surfaces without post-processing (10.3-30 μm). © 2021 Taylor & Francis.