Shape defect analysis from volumetric data - Application to lattice struts in additive manufacturing

Marc-Antoine de Pastre (1) , Yann Quinsat (1) , Claire Lartigue

1 LURPA - Laboratoire Universitaire de Recherche en Production Automatisée

Additive manufacturing (AM) revolutionises the way parts are produced as it offers a variety of design freedom. Lattice structures are an illustration of this AM freedom, allowing for the production of complex geometries that are being investigated in many applications. However, lattice structures present different typology of defects such as surface quality, porosity or dimensional inaccuracies. The most adapted measurement technology to reveal AM internal defects falls into X-ray computed tomography (XCT). Although there have been significant efforts in modelling lattice structure defects from XCT, a direct and accurate link between volumetric data of the part being measured and the CAD model is still required. This direct link would have the noteworthy advantage of not involving XCT surface determination tool, which choice may be discussed. In this paper, shape defects from metal laser powder bed fusion (PBF) strut-based lattice structures are studied. Different struts are printed as representative of BCCz lattice cells. Struts are successively measured by XCT and focus variation (FV). A virtual volume correlation (V2C) method is presented where shape defect contained in XCT volumetric data is successively approached by modal decomposition relying on a generated defect basis. The modal decomposition approach is firstly validated by comparing its efficiency towards least square cylinder approximation. Then, correlation intrinsic parameters are found, by conducting 2-dimensional sensitivity studies to identify optimal V2C parameters. V2C is further applied to the entire XCT measurements for each considered strut. Comparisons between correlated envelopes and registered FV and XCT measurements are performed to numerically estimate RMS errors. Results show that RMS errors between correlated envelopes and registered measurements are in the same order as the XCT resolution. Conclusions can then be drawn regarding the ability of V2C to estimate lattice strut shape defect relying on an user-defined shape defect basis.

Mots clés: computed tomography; virtual volume correlation; modal decomposition; Shape defect identification

Domaines: Génie mécanique [physics.class-ph]

Marc-Antoine de Pastre, Yann Quinsat, Claire Lartigue. Shape defect analysis from volumetric data - Application to lattice struts in additive manufacturing. Precision Engineering, 2022, 76, pp.12-28. ⟨10.1016/j.precisioneng.2022.02.011⟩. ⟨hal-03610605⟩