To permit bolting of a 90 lb/yd. flat-bottomed rail to a steel structure, rectangular slots 2 in. wide x 1 in. deep were flame-cut in the base of the rail at 2 ft intervals to suit existing bolt holes. During subsequent handling, one of the rails (which were about 25 ft long) was dropped from a height of approximately 6 ft on to a concrete floor and it fractured into 11 pieces, each break occurring at a slot. The sample piece submitted for examination showed a wholly brittle fracture at each end, the fractures having originated at the sharp corners of the slots. During flame-cutting, a narrow band of material on each side of the cut was raised above the hardening temperature. When the torch had passed the rate of abstraction of heat from this zone by conduction into the cold mass of the rail was sufficiently rapid to amount to a quench and thus cause local hardening. The steel in the regions of the slots possessed little capacity for deformation, and fracturing of the martensitic layer, under cooling or impact stresses, would be likely to occur. The slots should have been cut mechanically.

Several of the welds in a hoist carriage tram-rail assembly fabricated by shielded metal arc welding the leg of a large T-section 1020 steel beam to the leg of a smaller T-section 1050 steel rail failed in one portion of the assembly. Four weld cracks and several indefinite indications were found by magnetic-particle inspection. The cracks were revealed by metallographic examination to have originated in the HAZs in the rail section. Cracks in welds and in HAZs resulting from arcing the electrode adjacent to the weld and weld spatter were also revealed. The tram-rail assembly was concluded to have failed by fatigue cracking in HAZs. The fatigue cracking was initiated and propagated by vibration of the tram rail by movement of the hoist carriage on the rail. As a corrective measure, welding procedures were improved and the replacement rail assemblies were preheated and postheated.

A rail section that failed due to fatigue showed a smooth surface with well-developed conchoidal markings. This indicated successive stages of crack propagation, characteristic of fatigue failure. The crack was one of several which developed in the sections of curved rail which formed the lower roller path on which the superstructure of a walking drag-line excavator slewed. The cracking, which ran horizontally, developed at the junction of the underside of the rail head with the web and originated at surface defects in the form of grooves present on the castings. It was concluded that the cracking was caused by lateral deflection of the rails under in-service loads. The web of a rail would normally be loaded in compression but, should lateral movements occur, then it would be subjected to bending stresses and fatigue cracks could break out in regions where excessive tensile components predominated.

On 15 March 2000, a National Railroad Passenger Corporation (Amtrak) train traveling from Chicago to Los Angeles derailed in Carbondale, KS. After the initial on-scene investigation, 12 pieces of rail were sent to the materials laboratory for examination. Ten of them were from the point of derailment (POD). A vertical crack was observed in the head of the rail (vertical split head). The crack was at least 233 in. (591 cm) long, continuing through the entire lengths of most pieces recovered from the POD. The vertical fracture surface had features consistent with overstress fracture with short-term exposure to an oxygen-rich environment. Fracture features emanated from longitudinally-aligned inclusions rich in aluminum.