Statistical average strain energy density fatigue estimation of strut-based metamaterials via synthetic as-built CAD digital twins

The limited and scattered fatigue performances and their dif cult predictability remain critical barriers for the widespread adoption of Laser-based Powder Bed Fusion (L-PBF) metamaterials in engineering applications, as fatigue damage initiation is highly sensitive to manufacturing-induced geometric imperfections. While X-ray computed tomography (CT) provides high- delity as-built reconstructions fundamental for metamaterials’ structural health monitoring, its cost and complexity hinder routine integration into fatigue assessment work ows at the design stage. In this work, we propose a computationally ef cient framework for the development of synthetic as-built CAD models, serving as digital twins for fatigue life and failure location prediction. The proposed model is herein reported for L-PBF Ti-6Al-4V struts, the elemental building blocks of metamaterial architectures, manufactured at different building orientations. Leveraging stereomicroscopy input images, a modular reconstruction pipeline capturing orientation-dependent surface morphology and part ially fused particles allows the generation of as-built CAD models that retain the geometric variability governing fatigue behaviour, without reliance on volumetric imaging. Synthetic models are coupled with nite element analyses and a statistical strain energy density criterion to identify failure-critical locations. Validation against CT-derived counterparts demonstrates close morphological agreement and, since the design stage, the ability to estimate fatigue life and predict experimental failure locations within established scatter bands.