Increasing Part Accuracy in Additive Manufacturing Processes Using a k-d Tree Based Clustered Adaptive Layering

Abstract

Additive manufacturing (AM) is widely used in aerospace, automobile, and medical industries for building highly accurate parts using a layer by layer approach. The stereolithography (STL) file is the standard file format used in AM machines and approximates the three-dimensional (3D) model of parts using planar triangles. However, as the STL file is an approximation of the actual computer aided design (CAD) surface, the geometric errors in the final manufactured parts are pronounced, particularly in those parts with highly curved surfaces. If the part is built with the minimum uniform layer thickness allowed by the AM machine, the manufactured part will typically have the best quality, but this will also result in a considerable increase in build time. Therefore, as a compromise, the part can be built with variable layer thicknesses, i.e., using an adaptive layering technique, which will reduce the part build time while still reducing the part errors and satisfying the geometric tolerance callouts on the part. This paper describes a new approach of determining the variable slices using a 3D k-d tree method. The paper validates the proposed k-d tree based adaptive layering approach for three test parts and documents the results by comparing the volumetric, cylindricity, sphericity, and profile errors obtained from this approach with those obtained using a uniform slicing method. Since current AM machines are incapable of handling adaptive slicing approach directly, a “pseudo” grouped adaptive layering approach is also proposed here. This “clustered slicing” technique will enable the fabrication of a part in bands of varying slice thicknesses with each band having clusters of uniform slice thicknesses. The proposed k-d tree based adaptive slicing approach along with clustered slicing has been validated with simulations of the test parts of different shapes.