Multi-Axial Fatigue Analysis of Thermite Rail Welds
Abstract
This dissertation aims to examine the high-cycle fatigue behavior of thermite-welded railroad rails. In particular, the occurrence of fatigue defects in web and base regions of the rail is addressed, as frequently reported by field surveys. Fatigue life estimates are made by means of a multi-axial critical-plane fatigue algorithm that computes the fatigue damage based on the time history of stress tensors. The finite element method is used to analyze a full-scale replica of the wheel-track system comprising axle, wheel, thermite-welded rail, tie plates, and ties. The time-dependent rolling of the wheel and thermal stresses caused by seasonal temperature variations are incorporated into the finite element simulation. In addition, the effects of the track foundation stiffness and thermite weld geometry are explored.
Fatigue crack nucleation is studied at three critical locations of the rail, where most reported weld service failures occur: web-to-base fillet, base center, and base corners. Under wheel loads, the results of the fatigue analysis indicate that it can take a long period of time for a fatigue crack to nucleate in the aforementioned regions of the rail if the material is nominally “defect-free”. Fatigue cracks tend to initiate in a transverse plane, next to the thermite weld, perpendicular to flexural tensile stresses that form in the rail base. The implementation of small planar surface defects at critical locations has shown to significantly reduce the fatigue nucleation life of thermite rail welds. Also, axial tensile stresses resulting from cold winter weather are found to considerably accelerate the fatigue nucleation process, especially in the web-to-base fillet region.
Subject
Multi-Axial Fatigue AnalysisCritical-Plane Approach
Thermite Rail Welds
Finite Element Analysis of the Wheel-Track System
Weld Planar Defects
Rail Temperature Variations
Citation
Tavakoli, Maryam (2018). Multi-Axial Fatigue Analysis of Thermite Rail Welds. Doctoral dissertation, Texas A & M University. Available electronically from https : / /hdl .handle .net /1969 .1 /174627.