A high-temperature, high-cycle fatigue test was conducted on a nickel-based single-crystal superalloy with a pore structure. Optical and scanning electron microscopy were utilized to examine the crack propagation paths and fatigue fracture surfaces at the macro and micro scales. The analysis of crack initiation and propagation related to the pore structure facilitated the development of a crack shape factor reflecting these distinct fracture behaviors. Predictions about the high-cycle fatigue stress experienced by the specimen were made, accompanied by an error analysis, providing critical insights for precise stress calculations and structural optimization in engine blade design. The results reveal that high-cycle fatigue cracks originate from corner cracks at pore edges, with the initial propagation displaying smooth crystallographic plane features. Subsequent stages show clear fatigue arc patterns in the propagation zones. The fracture surface exhibits the significant layering of oxide layers, primarily composed of NiO, with traces of CoO displaying columnar growth. AL2O3 is predominantly found at the interfaces between the matrix and oxide layers. Short and straight dislocations near the oxide layers and within the matrix suggest that dislocation multiplication and planar slip dominate the slip mechanisms in this alloy. The orientation of the fracture surface is mainly perpendicular to the load direction, with minor inclined facets in localized areas. Correlations were established between the plastic zone dimensions at the crack tips and the corresponding fatigue stresses. Without grain boundaries in single-crystal alloys, these dimensions are easily derived as parameters for fatigue stress analysis. The selected crack shape factor, “elliptical corner crack at pore edges”, captures the initiation and propagation traits relevant to porous structures. Subsequent calculations, accounting for the impact of oxide layers on stress assessments, indicated an error ratio ranging from 1.00 to 1.21 compared to nominal stress values.
Read full abstract