Micro-CT Scanning in the Investigation of Squat Defects in Rail Steel

This thesis contains the results from an investigation into Squats, a discrete rail steel defect. Micro-computed tomography (micro-CT) scanning was used to scan the entire structure of four out of five defects removed from railway track. The fifth was not scanned as it shelled in track. The difficulty turning these raw X-ray images into a segmented 3D model were overcome by developing a new technique. Isolating separate regions of the scans created areas where voxel value variation was at a minimum (i.e. the histogram became one narrow peak rather than multiple broad peaks). This allowed the automatic crack segregation module to work fairly well, and then enhanced using the region growing modules within the software. These scan results were then verified to be accurate using metallographic sample preparation, optical and electron microscopy.
Micro CT scanning was performed on a custom 450KeV scanner, allowing the capture of the first entire Squat crack network morphology. Full defect imaging allowed the different defects to be compared to each other, highlighting differences and similarities. The defects came from metro, mixed and high-speed railways and one was found within an aluminothermic weld. The verification process and investigations of the scan volumes yielded further information about the defect’s origins. These origins were used to determine that, for the five defects investigated, there were four different causes. The two that shared a cause were from the same track section. Based on the causes, the defects were identified as a Stud, two Grinding Induced Squats (GIS), a Squat (caused by the legacy issue of MnS inclusions) and possibly a Squat or Stud in a slightly contaminated weld. None of the defects were considered to be a classic Squat, which is caused by Rolling Contact Fatigue (RCF), because there were other factors in their initiation.
One of the defects contained a transverse defect, which is a crack that grows down through the railhead and can break the rail. This transverse defect was ~9mm deep into the rail when it was removed, meaning it would not have returned to the surface to shell. A high-resolution volume of the transverse defect region was created and its origin gives an important insight into the potential causes of rail-breaking defects. The origin of this transverse defect was a cluster of debris-filled voids that had formed due to the corrosion and cyclic loading (fretting) of a crack branch. These voids aligned with a deep grinding mark on the surface of the rail, which acted as a stress raiser. Because corrosion is a factor in this transverse defect case, the age of a rail and its environment are factors for defect development as well as traffic volume, given the correlation of corrosion with time.
Results of this work highlight both the importance of a good surface finish and the diversity of causes found within the term “Squat”. Thus the identification of the Stud variant may be the beginning of a more comprehensive group of Squat type defects being established. This refining of the category could lead to fruitful big data analyses of the Squat type defect occurrences.
The CT volumes of the defects created in this work can easily be stored for comparison in future investigations. The virtual nature of the volumes allows the sharing of defect information more readily than physical and sectioned defects, which deteriorate with time and require physical storage and transport.