Mechanical properties and fatigue resistance of rails welded with the aluminothermic method

CIM Bulletin, Vol. 2, No. 4, 2007

J.M. Duart, J.O. García, J.I. Verdeja, and J.P. Sancho

The aluminothermic welding process is based on the highly exothermic reaction between iron oxide, alloying elements, and aluminium. The rails are previously preheated (900ºC) with skill to avoid the martensite-bainite structure from cooling quickly. Welding uses the KLK Spanish technology kit. European Community regulations on standard tests to guarantee weld quality are: Hardness test 10/3,000 (250 ± 20 to 350 ± 20 HBW. Fully pearlitic rail structure. Martensite is not accepted. The slow bend test gives the minimum load and the deflections allowed, according to the standards and profiles of the rail (see table). Welds must support, as a minimum, an equivalent stress of around 60% of the rail’s tensile strength. The weld rail sample must withstand a minimum fracture load of 750 MPa in tension. Fatigue produced by a repeated charge-discharge, simulating use, determines track and weld life. Observing shafts and wheel failures, Wöhler developed the standard test for fatigue determination and established the Wöhler laws. The life of a material may be subject to statistical effects, calculable in occasions. Rail breakdown by fatigue can only be avoided by the preventive periodical replacement of track sections. Rail cracks can be: head cracks in the form of an oval spot (stain), of easy detection track usually breaks transversally; foot cracks with thumb-nail form, which are dark in colour, much more dangerous than the former; web to foot cracks, with parabolic contours. The track should support 10*106 cycles of charge-discharge in all its welded rail-weld-rail unions. The standard fatigue test uses a 1,000 mm distance between supports, a load on two rolls spaced 50 mm. The slow bend test reveals gross defects in manufacturing: skin defects, macro and micro shrink cavities, segregations, porosity due to evolving gases in the process. Fatigue range arises from the experience of the statistical nature of the Wöhler test. The scattered points around a middle line are usually highlighted and fit a normal distribution. The Wöhler curve defined would include the range m ± s in which 68% of fatigue fractures should occur (m = fatigue limit and s = standard deviation, from the Wöhler curve). Fatigue limit at 2*106 cycles is: m = 275 MPa; standard deviation, s = ±19. The Locati method is preferred to the staircase method because it is simple, quick-operating, and very suitable for quality control of the welding process. By using this method, the following can be obtained: smf = 270 MPa, at 2*106 cycles. Conclusions The aluminothermic kits investigated with the short preheat technology process are easy, safe, and fulfil the E.C. railroad regulations on aluminothermic welding. The welding process gives the homogeneity required by the most common rails, and has a similar hardness, soundness, and correct microstructure to that of the rails without martensite and bainite. They resist loads and allow greater strains as measured by the slow bend test and fatigue test required by railways. The fatigue test of aluminothermic welded rails is highly suitable for the internal and external soundness—demonstration of the welded joint. The 50% fatigue limit and its standard deviation are 234 ± 19 MPa at 2*106 cycles, however, doing eight tests. The Locati method, with only one test, gives values (m = 270 MPa) very similar to those of the staircase method (E.C. standards). The most frequent causes of fracture and fatigue in the aluminothermic welded rails tested are associated to a defective design of the weld collar web-foot union zone, soft-skinned zones with thickness greater than 0.5 mm, lack of fusion between rail-foot and the weld collar, and external defects in the lower rail-foot weld collar.