Simulation-based failure analysis of faulty and regulatory railhead repair welding procedures

Abstract

Advancements in railway technology have significantly reduced wheel and axle failures, yet rail failures, particularly in welded regions, remain a major concern. Up to 60% of recorded rail failures occur in these areas, with accident reports frequently attributing incidents to improperly executed repair welds. Using thermo-metallurgical-mechanical finite element simulations, this study investigates the mechanical performance of regulatory and faulty in-situ railhead repair welds, where only part of the railhead is removed. The multi-pass welding simulations employ a multi-phase homogenization-based material model, incorporating non-linear isotropic and kinematic hardening, phase transformation kinetics, and virgin material state recovery. Mechanical performance of the railhead repairs is evaluated through wheel-rail contact over-rolling simulations and fatigue analysis using novel multi-phase Dang Van criterion.

The comparison, based on quantitative data, reveals an increased risk of fatigue crack initiation when deviations from the regulatory procedure occur. The regulatory repair produces a ferritic-pearlitic microstructure and a more favorable residual stress state, characterized by compressive stresses in the weld region and tensile stresses deeper below the rail surface where the microstructure is more favorable. In contrast, the faulty repair exhibits rapid cooling rates, leading to brittle martensitic phases and high tensile residual stresses near the rail surface, significantly increasing fatigue crack initiation risk. Operational over-rolling simulations further demonstrate that tensile stresses in the faulty repair persist near the surface, failing to redistribute as effectively as in the regulatory repair. These findings underscore the importance of strict adherence to repair welding standards to prevent premature rail failures and costly maintenance interventions.