Microstructure-sensitive crystal plasticity and fatigue indicator modeling for LZ50 steel

The fatigue performance of railway axle steel is highly sensitive to microstructural heterogeneities and internal defects, which are inadequately captured by conventional life prediction methods. Motivated by this, a two-stage fatigue life prediction framework for LZ50 steel is employed that integrates the crystal plasticity finite element method with fatigue indicator parameters to account for microstructure-sensitive fatigue processes, including crack initiation and microstructurally short crack growth. To establish a typical experimental foundation, microstructural characterization via electron backscatter diffraction and scanning electron microscopy, displacement-controlled uniaxial tensile tests, and strain-controlled fatigue experiments were conducted. Representative volume elements were constructed based on the characterized microstructures, and crystal plasticity parameters were calibrated against both tensile and fatigue test results obtained at a strain amplitude of 0.9%, and further validated at amplitudes of 0.45%, 0.6%, and 0.75%. Compared to the approaches based on conventional fatigue indicator parameters, the two-stage framework that decouples crack initiation and microstructurally short-crack growth significantly improves prediction accuracy, with all results falling within the ± 1 . 5 × scatter band. The microstructurally short crack growth stage is found to contribute more than 50% of the total fatigue life. Furthermore, the effects of inclusions and pores with varying size, shape, and stiffness are systematically investigated. This study provides an effective and physically grounded framework for fatigue life prediction of defect-containing microstructures in structural steels. • A CPFEM–based framework is developed for fatigue life prediction of LZ50 axle steel. • Three types of FIPs are evaluated and a new FIP-based MSC propagation criterion is proposed. • Realistic dual-phase RVEs are generated and calibrated against experimental data. • The proposed method achieves high prediction accuracy across multiple strain amplitudes. • The effects of inclusions and pores on fatigue crack initiation are systematically investigated.