Geometric deviations of laser powder bed–fused AlSi10Mg components: numerical predictions versus experimental measurements

February 28, 2020

Floriane Zongo (1), Charles Simoneau (2), Anatolie Timercan (1), Antoine Tahan (1), Vladimir Brailovski (1)
The International Journal of Advanced Manufacturing Technology, 107, February 2020: 1411–1436. DOI: 10.1007/s00170-020-04987-7


Laser powder bed fusion, Geometrical deviations, Numerical predictions, Predictions capabilities


Laser powder bed fusion (LPBF) is one of the most potent additive manufacturing processes. One of the constraints for a broader industrial use of this process is the limited knowledge of its dimensional performances and geometrical behavior, as well as the inability to predict them as a function of material, process parameters, part size, and geometry. The objective of this study is to enrich knowledge of the geometric dimensioning and tolerancing (GD&T) performances of the LPBF process and to evaluate the distortion prediction capabilities of the ANSYS Additive Print® software. To this end, a selected topologically optimized part with three different support configurations was manufactured using an EOSINT M280 printer and AlSi10Mg powder. After printing, the parts were scanned using a coordinate measuring machine (CMM) and a micro-computed tomography (μ-CT) system. The GD&T calculations were carried out according to the ASME Y14.5 (2009) standard. The distortions measured by the CMM and μ-CT techniques were 0.195 mm and 0.368 mm, respectively (95% interval). After the software calibration and two numerical sensitivity studies, the same stereolithography files used to print the parts were downloaded into the ANSYS Additive Print® software to calculate distortions caused by the process. The differences between the experimentally measured and the ANSYS-predicted distortions for a 56 mm × 58 mm × 137 mm part fell within a 0.134 mm range at a 95% interval. The fidelity of the numerical predictions, the impact of the support structures, and the differences induced by the CMM and μ-CT measurement uncertainties are presented and discussed.

How Our Software Was Used

Dragonfly was used to separate the region of interest (ROI) from the void in CT-scan data.

Author Affiliation

(1) École de technologie supérieure, Montreal, Quebec, Canada.
(2) SimuTech Group, Montreal, Quebec, Canada.