Abstract
In many additive manufacturing processes, surface roughness is a critical quality concern. Due to the nature of the layer-by-layer manufacturing process, the pattern of surface roughness depends on the location on the surface, i.e., the layer number and the location within each layer. Adequate description of the surface roughness enables us to develop effective post-processing plans, reveal the root causes of the roughness, and generate accurate compensation schemes.
In this work, we propose a three-step surface roughness characterization method (SRCM). This method is based on the dense point cloud data generated from the surface scan of additively manufactured products. First, we use a double kernel smoothing spatial variogram estimator to represent the heterogeneous roughness property at different surface locations. Second, we extract the magnitude and scale of surface roughness from the estimated variogram. Third, we use Gaussian Process to build a roughness map on the entire surface based on the roughness characterization on these sampled points.
The SRCM is demonstrated from a high-density 3D scan of a cylindrical product fabricated by a wire-arc additive manufacturing process. It shows that our approach serves as an effective tool to infer the roughness map from the 3D point cloud data. In the end, we will briefly discuss how to use the inferred roughness map to develop an optimal surface smoothing method.
