Achieving rapid and uniform deep austenitization in rails via variable-speed induction heating for gradient microstructure control

Achieving rapid, uniform, and deep austenitization in large, complex-shaped workpieces like rails is critical for gradient microstructure control, yet remains challenging because of the inherent temperature non-uniformity in conventional induction heating. To address this challenge, this study proposes an innovative variable-speed moving induction heating strategy using a profiling coil, designed to meet the localized heating demands of the rail geometry. A 3D coupled electromagnetic-thermal-motion finite element model was established and experimentally validated to analyze temperature distribution characteristics during moving induction heating. Key process parameters, including current density, current frequency, and moving speed, were systematically optimized to achieve rapid and uniform austenitization via simulation. The results demonstrated that increasing current density and frequency elevated heating efficiency but compromised temperature uniformity. Adjusting the moving speed alleviated non-uniformity but could cause surface overheating. By employing a variable-speed induction heating process, excessive surface heat accumulation was effectively reduced, yielding a more homogeneous temperature field. The optimized approach achieved full austenitization within the top 15 mm of the rail in only 10 s and improved temperature uniformity by 22.4%. The integrated simulation-experimental methodology and variable-speed control logic established a generalizable framework for the precise localized heating of complex-shaped industrial workpieces. To further validate the proposed process, thermal history simulation, microstructural observation, and hardness testing were performed at different depths. The results establish a complete thermal–microstructure–property relationship, confirming that the optimized thermal field can induce a controllable gradient microstructural response, providing a reliable basis for gradient property control.