Hybrid Phenomenological-Empirical Modeling of Catalyst Deactivation in an Industrial Steam Methane Reformer

Industrial steam methane reforming remains the dominant large-scale route for hydrogen production, yet long-term catalyst deactivation and tube-wall temperature evolution are rarely incorporated simultaneously into industrial reactor models. This work presents a hybrid phenomenological–empirical model for an industrial top-fired steam methane reformer that combines a one-dimensional pseudo-homogeneous reactor model, a catalyst deactivation model with residual activity, and an empirical wall-temperature model based on routinely measured process variables. Model parameters were estimated simultaneously from more than 1000 days of industrial operating data, using a hybrid global–local optimization strategy. Parameter uncertainty was then assessed using residual bootstrap. The proposed framework accurately reproduced methane conversion and reformer outlet temperature, with mean absolute percentage errors below 5%, negligible systematic bias in outlet temperature predictions, and only a modest systematic underprediction of −2.00% in methane conversion throughout the operating campaign. The estimated deactivation parameters were consistent with sintering-related aging mechanisms reported for Ni-based steam reforming catalysts. Despite its reduced complexity, the model successfully captured the coupled evolution of catalyst activity and furnace thermal behavior over nearly three years of industrial operation. The proposed methodology provides a practical and computationally efficient framework for long-term process monitoring, catalyst health assessment, campaign analysis, and predictive maintenance of industrial steam methane reformers.