Hot-dip galvanizing lines require precise control of surface decarburization to achieve target mechanical properties in automotive steel sheets. This study presents a coupled surface-reaction and one-dimensional carbon diffusion model for predicting decarburization during non-isothermal annealing. The model incorporates thermodynamic equilibrium phase fractions from the Fe–C system to describe ferrite–austenite distribution as a function of local temperature and composition, with an effective diffusion coefficient that transitions smoothly between phase-specific diffusivities using a phase-transition parameter. Model parameters were optimized using differential evolution against Glow Discharge Optical Emission Spectroscopy measurements from nine annealing experiments at \({750}\,^{\circ }\) C to \({850}\,^{\circ }\) C with varying \( {p_{{\text{H}}_2{\text{O}}}}/{p_{{\text{H}}_2}}\) ratios (0.0077 to 0.1268). Leave-one-out cross-validation yielded a symmetric mean absolute percentage error of 14.6 pct [95 pct CI: 10.0 to 19.3 pct], while in-sample fitting achieved 10.6 pct. Statistical analysis revealed that decarburization depth was primarily influenced by the \( {p_{{\text{H}}_2{\text{O}}}}/{p_{{\text{H}}_2}}\) ratio, with soaking temperature showing only a slight effect within the investigated range. The model successfully captures temporal evolution of carbon concentration profiles during heating and cooling cycles, providing insights into diffusion-limited regimes at elevated temperatures. Limitations include exclusion of oxidation effects and indirect treatment of phase transformation kinetics through empirical phase-transition parameters. This framework enables cost-effective prediction of decarburization under various process conditions while identifying key atmospheric parameters for process optimization in hot-dip galvanizing lines.
