Multiscale and multitechnique investigation of the elasticity of grooved rail steel

Abstract

Tramway rail steel is exposed to extreme temperature conditions both during production (e.g. in terms of heat treatment) and over its decades-to-century-long service life (e.g. in terms of welding operations in the course of maintenance). The question arises whether this induces local stiffness reductions and hence stiffness inhomogeneities at the half-millimeter level; i.e. the characteristic size governing the structural behavior of the rail. In order to address this question, a series of 16800 nanoindentation tests, with indentation depths ranging from 200 to 250 nm (characterizing 66- to 125 nm-sized material volumes), were performed on samples with less than 10 nm surface roughness, extracted from different locations of typical tramway rail cross sections at different time points during their service lives, and including both heat-affected and non-heat-affected zones. Thereby, each of these locations was probed through a grid of 20 x 20 nanoindentations, with a grid spacing of 500 μm. In very few cases, the indentation tip was probably moved into cracks of several μm width and tens-to-hundreds of μm length; as seen on light and electron micrographs. Zero-stiffness was assigned to the corresponding grid points. The probability distributions of the remaining, non-zero elastic moduli were fitted by one to four weighted Gaussians, representing all (non-zero) nanoindentation data as well as particular data subsets (namely data from each testing grid, data from each testing location across all rails, data from each rail, all data from heat-affected zones, and all data from non-heat-affected zones). The aforementioned Gaussians refer to different solid elastic material phases, with expected Young’s moduli ranging approximately from 100 to around 300 GPa, potentially reflecting different dislocation densities. On the other hand, at a larger material scale (i.e., that of 220 to 440 μm, tested through 2.25 MHz ultrasonics), the stiffness is reduced to approximately 213 GPa (from a mean nanoindentation-derived value of approximately 246 GPa). This reduction can be explained by a micromechanical model, with the intact steel stiffness as well as with crack sizes and crack numbers as input quantities.