Damage assessment in a wheel steel under dry-lubricated contact by an innovative vision system

The alternation of dry and wet contact can cause severe damage on railway wheels, because ratcheting occurring in dry contact leads to the formation of surface cracks, which can propagate if water is subsequently added, owing to the entrapped fluid pressurization mechanism. In a previous work, damage in alternated dry-wet contact was studied by means of twin-disc tests, aided by an innovative vision system including two devices: one for assessing the surface status by elaborating lights and shadows due to a diffused laser light projected on the specimen surface (2D analysis), the other for quantifying the local damage depth by elaborating the deformations of a projected laser blade (3D analysis). In that work, the 2D analysis resulted effective in revealing the onset of diffused fatigue crack propagation before it could be detected by vibration increment or visible spalling, but it failed when the damage was local; on the contrary, 3D analysis was effective in evaluating the damage severity only when the final failure had occurred. In this paper, the 2D analysis method was improved, including localised analyses in addition to the analysis of the whole contact surface. Two tests were considered: one with alternated dry-wet contact sessions of 50000 cycles each, the other with a one million cycle dry session followed by wet contact up to failure. The 2D analysis results were compared with the 3D analysis ones; furthermore, at the end of the tests the specimens were cut at the locations where the vision systems indicated severe damage. In the test with short alternated dry-wet sessions the local 2D analysis results showed the same trend as the global ones: indeed, in this test the damage was diffused along the specimen surface, the final failure resulting in many spalls all over the surface. Comparing the results of the 2D and 3D vision analyses with the observed crack depth, a general agreement was found. In the test with the long dry contact session before the wet one, an agreement between the local 2D and 3D again was found only if the 2D analysis was performed locally, as the failure was determined by few large spalls. Again, a general agreement was found between the 2D analysis results and the observed subsurface crack depth. The improvement of the 2D analysis procedure showed therefore the effectiveness of this method as an on-line non-destructive diagnostic tool even in case of localised damage.