Cost-efficient testing of highly loaded fatigue cracks in aerospace alloys under realistic fuselage loading conditions supported by intelligent algorithms

Accurate lifetime estimations of lightweight aircraft structures are crucial for safety and economic reasons. Fatigue cracks are likely to occur in the fuselage owing to non-constant service loads, wall thickness below 2 mm and stresses higher than 100 MPa. The presence of such cracks is included in the design of aircraft structures but their behaviour must be well understood to implement reliable damage tolerance concepts. In this work uniaxial and biaxial tests are used to study propagation of fatigue crack with lengths > 300 mm and stress intensity factors of over 100 MPa√m in rolled AA2024 and AA5028 sheets with thickness of 1.6 mm and 1.8 mm. Large biaxial and uniaxial tests specimens with dimensions of 2098 x 1734 mm² and 1290 x 950 mm² were investigated. The tests were supported by an in situ 3D digital image correlation (DIC) system to obtain full-field strain and displacement information on the surface during crack propagation. A DIC-based system was developed to determine the crack path with a pre-trained convolutional neuronal network. The results obtained were then used to compute the actual crack tip loading K_I/K_II with the help of a post-processor based on the interaction integral. The crack propagation curves, da/dN-ΔK, could be determined with stable crack propagation rates of more than 1 mm/cycle and ΔK > 100 MPa√m. The combination of classical testing methods with digital evaluation algorithms presented in this work enable new insights into the fracture mechanics behaviour of large fatigue cracks.