Development and Validation of Empirical Models to Predict Metal Additively Manufactured Part Density and Surface Roughness from Powder Characteristics

Abstract

Metal additive manufacturing (AM) processes, viz laser powder bed fusion (L-PBF), are becoming an increasingly popular manufacturing tool for a range of industries. The powder material used in L-PBF is costly, and it is rare for a single batch of powder to be used in a single L-PBF build. The un-melted powder material can be sieved and recycled for further builds, significantly increasing its utilisation. Previous studies conducted by the authors have tracked the effect of both powder recycling and powder rejuvenation processes on the powder characteristics and L-PBF part properties. This paper investigates the use of multiple linear regression to build empirical models to predict the part density and surface roughness of 316L stainless steel parts manufactured using recycled and rejuvenated powder based on the powder characteristics. The developed models built on the understanding of the effect of powder characteristics on the part properties. The developed models were found to be capable of predicting the part density and surface roughness to within ±0.02% and ±0.5 Ra, respectively. The models developed enable L-PBF operators to input powder characteristics and predict the expected part density and surface roughness.