Additive manufacturing is an emerging technology and provides high design flexibility to customers. Fused deposition modeling (FDM) is an economical and promising additive manufacturing method. Due to its many advantages, FDM received great attention in recent years, and comprehensive studies are being undertaken to investigate the properties of FDM-printed polymers and polymer composites. As a result of the manufacturing technology employed in FDM, inner structures are changed with different process parameters, and thus, anisotropic properties are observed. Moreover, composite filaments such as particle- or fiber-reinforced polymers already have anisotropy before FDM printing. In this study, we investigate the effect of different process parameters, namely layer thickness and raster width on FDM-printed copper-reinforced poly(lactic acid) (PLA). Mechanical characterizations with a high-resolution camera are carried out for analyzing the deformation behaviors. Optical microscopy characterizations are performed to observe the mesostructural changes with various process parameters. Scanning electron microscopy (SEM) and an energy-dispersive X-ray spectroscopy (EDS) analysis are conducted for investigating the microstructure, specifically, copper particles in the PLA matrix. A 2D digital image correlation code with a machine learning algorithm is applied to the optical characterization and SEM-EDS images. In this way, micro- and mesostructural features, as well as the porosity ratios of the specimens are investigated. We prepare the multiscale homogenization by finite element method (FEM) simulations to capture the material's response, both on a microscale and a mesoscale. We determined that the mesostructure and, thereby, the mechanical properties are significantly changed with the aforementioned process parameters. A lower layer thickness and a greater raster width led to a higher elasticity modulus and ultimate tensile strength (UTS). The optical microscopy analysis verified this statement: Decreasing the layer thickness and increasing the raster width result in larger contact lines between adjacent layers and, hence, lower porosity on the mesoscale. Realistic CAD images were prepared regarding the mesostructural differences and porosity ratios. Ultimately, all these changes are accurately modeled with mesoscale and multiscale simulations. The simulation results are validated by laboratory experiments.
Keywords: additive manufacturing; digital image correlation; finite element method; fused deposition modeling (FDM); machine learning; multiscale homogenization; polymer composites.