Quantifying the effect of process-induced defects on the compressive response of 3D-printed lattices

Abstract This study investigates the influence of selective laser sintering (SLS)-induced defects on the quasi-static compressive response of PA2200 body-centred cubic (BCC) lattice structures using a combined experimental and numerical approach. Defect-informed finite element models were developed by incorporating statistically characterised geometry-related imperfections, including strut under-sizing, cross-sectional distortion, and thickness variation, based on previously characterised SLS-manufactured lattices produced under the same processing conditions. To separate the influence of parent-material behaviour from the influence of manufacturing defects, defect-free compression-moulded PA2200 specimens were produced from the same powder system and thermally conditioned to achieve a degree of crystallinity close to that of the printed lattices. The predicted compressive responses of the geometry-defect-inclusive lattice models were compared with quasi-static compression experiments on printed BCC lattice specimens. The results show that incorporating realistic geometric imperfections substantially improves agreement with the measured response compared with an ideal-geometry model. Within the tested material, topology, and processing condition, strut under-sizing was found to have a pronounced influence on lattice performance: a 12% reduction in strut thickness led to an approximately 40% reduction in Young’s modulus, yield stress, compressive strength, and plateau stress. Unit-cell simulations further suggest that non-spherical internal voids have a smaller influence on stiffness and strength than geometry-related defects for the void volume fractions considered. Overall, the results indicate that defect-informed modelling can support more accurate interpretation of the compressive behaviour of SLS-manufactured polymer lattices. Further validation across additional lattice topologies, build conditions, relative densities, and loading modes is required before the framework can be generalised for broader process optimisation or design compensation.