Abstract Additive manufacturing (AM) is a transformative technology that enables the fabrication of complex geometries layer by layer. However, metal parts produced via AM processes such as laser powder bed fusion (LPBF) are prone to various defects, including porosity and deformation. These defects often result from suboptimal printing parameter settings. Traditional approaches typically aim to reduce defects by optimizing a fixed set of parameters for the entire part. However, such methods do not account for layer-wise variations in printing conditions caused by changes in geometry, heat transfer, and re-heating effects. While optimizing parameters for each layer could improve part quality, it would require an impractically large number of experiments for parts composed of hundreds or thousands of layers. To address this challenge, this paper proposes a novel approach in which printing parameters along the build direction are modeled using a parameter function. This function is constructed as a weighted combination of several basis functions, substantially reducing the number of decision variables. The weights are optimized using a response surface method to minimize a defect index, which aggregates multiple quality metrics, including residual stress, displacement, lack of fusion, and overheating. The proposed method significantly reduces porosity in an inverted-cone geometry compared to conventional approaches, as demonstrated through finite element analysis simulations and physical experiments that optimize laser power. This research presents a robust framework for efficiently optimizing layer-wise printing parameters to minimize process-induced defects in LPBF.
Dou et al. (Fri,) studied this question.