Influence of manufacturing defaults on the behavior of 3D printed lattice structures with a multiscale data-driven approach

The inherent multiscale (MS) complexity of lattice materials necessitates specialized modeling techniques to predict their mechanical behavior accurately and efficiently. A combination of the multiscale finite element method (e.g. FE²), with a Data-Driven paradigm, leads to Multi Scale Data-Driven (MSDD) modeling procedure to simulate the behavior of these materials. On top of the inherent complexity of these materials, limitations of the Laser Powder Bed Fusion (LPBF) manufacturing process result in geometrical imperfections of the manufactured materials giving rise to so called as-manufactured configuration. Furthermore, theses imperfections have significant influence on the mechanical behavior and introduce biases and variance from expected behavior based on the modeling with initially designed lattice material, so called as-designed configuration. The proposed research is the first step in the reduction of the cost of MSDD approach by aligning the model's accuracy with the manufacturing capabilities. The aim is to exploit the inherent variance in as-manufactured lattice material and to minimizing the material database to the essential material state points. To that end, in this work, using X-ray tomography, we create digital twins of various as-manufactured aluminum lattice unit cells and perform a number of simulations. The goal is to get further understanding of the variance in mechanical behavior resulting from geometrical variations and defects. These results will be further used to enhance the efficiency of the MSDD approach with no compromise on predictive capabilities.