The influence of constraint on fitting fatigue crack growth data

Fracture mechanics, in terms of crack propagation, is one of the most popular approaches for life prediction of components and structures subjected to fatigue loads. Experimental tests are carried out in order to quantify crack growth rates and thresholds. The obtained results, together with dedicated numerical analyses, are used in the calibration process of crack propagation analytical models. These models are included in widespread life prediction softwares, such as AFGROW and NASGRO, which adopt, as propagation law, the so-called “NASGRO Equation” in which the constraint at the crack tip is kept into account by a “constraint factor”, initially proposed by Newman together with the Strip-Yield model. These existing softwares incorporate numerical procedures for fitting the parameters of the NASGRO Equation by using experimental da/dN data without any explicit reference to test conditions. On the other hand, fatigue crack growth data obtained with different experimental conditions (specimen type, maximum applied stress and stress ratio) are also differing in terms of constraint at the crack tip and, consequently, the direct fitting of the NASGRO Equation is not correct. In the present paper, a set of crack growth data, obtained onto a mild structural steel using different specimens (SE(B) and C(T)) is examined from the point of view of the constraint factor at the crack tip by means of detailed FEM analyses. Results permitted to show how apparently conflicting data (especially in terms of thresholds) can be re-analysed considering the constraint factors of the different tests and, consequently, how different life predictions can be.