A dimensional compensation algorithm for vertical bending deformation of 3D printed parts in selective laser sintering

Purpose The purpose of this study is to mitigate the dimensional inaccuracy due to vertical curling/bending deformation of three-dimensional (3D) printed parts produced by selective laser sintering (SLS) using PA12 based on dimensional compensation of the computer-aided design (CAD) model. Design/methodology/approach To carry out this study, specially designed features are initially produced as references, and the dimensional deviations from the vertical bending deformation of the SLS process are analyzed. Next, the deformation patterns are formulated using a polynomial regression model in the global Cartesian coordinates of the building platform. Then, the compensation algorithm is implemented and the original 3D CAD file is preprocessed with an inverse transformation of the features to compensate the deformation errors. Findings It was found that the 3D printed parts from the SLS process have the dimensional inaccuracy due to the vertical bending pattern of the quadratic form. By implementing the compensation algorithm, it was statistically shown to effectively reduce bending deformations of various sample parts, including the automotive components, in SLS. Research limitations/implications The position of samples in a batch has a direct impact on not only bending deformation but also on horizontal shape geometry error. However, the application of this algorithm is focused on the vertical bending deformation, which is estimated as a major part of dimensional inaccuracy. Practical implications This paper provides a practical case study with a real vehicle part. The algorithm was shown to provide a more realistic solution to the dimensional deformation of printed products, which is not manageable by simply using the constant scale factors provided by SLS 3D printer manufacturers. Originality/value This paper suggests that the vertical bending deformation from SLS’s 3D printed complex parts can be improved through the proposed compensation algorithm. The compensation algorithm was constructed by using the predictive regression model created from the bending deformation patterns of reference samples. The proposed compensation algorithm can be further used and applied for other complex samples without making additional reference parts.