This paper uses fracture mechanics to predict the fracture condition of squats, which are geometrically complex, multi-axially loaded fatigue cracks in rail. A squat consists of multiple non-planar fracture surfaces, leading to a challenge in transferring the fracture toughness obtained from small-scale specimens to the fracture condition of a squat. Therefore, in this study, a transferability framework is developed. The mixed mode fracture toughness of R260Mn rail steel is experimentally obtained using compact tension shear specimens. An equivalent fracture toughness based on the strain energy release rate was found to best describe mixed mode fracture toughness. Next to this, full-scale fracture tests are performed on rails of the same steels that include squats resulting from in-service loading. After fracture, the fatigue regions of the fracture surfaces are digitalized using a laser line scanner. The scanned surfaces are analysed in a numerical model. The extended finite element method is used to calculate the stress intensity factor distribution along the crack front. Considering the effects of crack front length and crack tip constraint, the stress intensity factor distribution is converted to a scalar failure condition. The calculated fracture conditions from the full-scale tests (average 48 MPa m with a coefficient of variation of 0.18) show good agreement with the experimentally obtained equivalent mixed mode fracture toughness (average 50 MPa m with a coefficient of variation of 0.1). To the knowledge of the authors, this is the first successful attempt to predict the fracture condition of squats, which are cracks in rails with complex non-planar fracture surfaces. • Squats are geometrically complex, multi-axially loaded crack networks in rails. • Fracture condition of in-service squats is estimated via a transferability framework. • The developed framework considers crack-tip constraint and statistical size effect. • Mixed mode fracture toughness of R260Mn is obtained with CTS tests. • The fracture condition of the squats agrees well with small-scale test results.
Hengeveld et al. (Wed,) studied this question.