Fatigue performance and fracture-based modeling of LPBF CuCrZr with process-induced defects

This study investigates the fatigue performance of laser powder bed fused (LPBF) CuCrZr alloy intentionally processed under non-optimized conditions to introduce lack-of-fusion (LOF) defects and coarse precipitates, enabling the development of defect-sensitive fatigue models. Specimens were produced with high LOF porosity and subjected to heat treatment to reproduce industrial-scale constraints. Tensile and fatigue tests, conducted on both smooth and notched specimens, revealed LOF pores, in subsurface or in touch with surface position, as the dominant crack initiation sites. A comprehensive characterization of pore size and distribution enabled the construction of a defect-inclusive Kitagawa–Takahashi diagram using the Benedetti–Santus semi-analytical approach, eliminating the need for costly fatigue crack growth threshold experiments. Five defect-tolerant fatigue prediction methods were benchmarked: Murakami's √area model, the extended Kitagawa–Takahashi framework, the Theory of Critical Distances (TCD), the Averaged Strain Energy Density (ASED) method, and the Bayesian Fatigue Analysis with Defects (B-FADE). The Benedetti KT model and B-FADE yielded accurate, experimentally validated fatigue thresholds, while TCD and ASED provided consistent defect-free fatigue limits and notch sensitivity indices (q ≈ 0.5). Murakami's model overestimated endurance limits due to its calibration on harder alloys. These findings emphasize the need for alloy-specific modeling strategies and demonstrate that accurate fatigue-life predictions for LPBF CuCrZr are achievable, even in the presence of severe processing-induced defects. This work supports the development of robust, defect-tolerant design approaches for high-performance AM copper components in aerospace, nuclear, and electronics applications. • Defect-tolerant design comparison by testing LPBF CuCrZr with high LOF porosity. • Defect-inclusive Kitagawa–Takahashi diagram without costly crack growth tests. • Benedetti–Santus model and Bayesian B-FADE accurately predict fatigue thresholds. • Murakami's √area model overestimates CuCrZr endurance limit by ∼30%. • TCD and ASED methods capture notch sensitivity (q ≈ 0.5) in coarse-grained Cu.