Modeling and experimental characterization of pearlitic rail steels subjected to large biaxial strains

Licentiate thesis, 2017

Large shear strains develop in the near-surface region under the running band of railway rails. Rolling Contact Fatigue (RCF) cracks often initiate in this region, causing major problems for the railway industry. However, characterization of the constitutive and fatigue behavior of this region is difficult due to the large gradient of properties. In the present thesis, the deformed microstructure in this region is characterized. An axial-torsion test rig is used to predeform cylindrical low-cycle fatigue specimens in order to obtain material properties similar to those of the near-surface region in rails. These specimens are more suitable for further mechanical testing, compared to those resulting from many of the other predeformation methods described in the literature. The obtained material is compared to field samples in terms of the material hardness and microstructure. The microstructure is evaluated with both optical microscopy and scanning electron microscopy. This comparison shows that the predeformed material state closely resembles what is found in some used rails at a depth between 50 and 100 μm. In order to describe the behavior of the material during the large shear deformations, a sound framework for finite strain metal plasticity is needed. Several options are available in the literature, but in this thesis two frameworks for hyperelasto-plasticity with kinematic hardening are investigated. It is shown that for appropriate choices of Helmholtz' free energy these frameworks are equivalent. Furthermore, several material models formulated within this framework are evaluated in terms of their abilities to predict the mechanical response during the predeformation. Particular emphasis is put on the role of the kinematic hardening laws and how these influence the response during the biaxial loading. It is found that by combining different models from the literature, the predeformation process can be modeled accurately.