Steam-iron process kinetic model using integral data regression

A kinetic model describing the gas–solid non-catalytic reaction between iron oxides and hydrogen/methane gas mixtures has been proposed. This steam-iron process constitutes an interesting alternative in order to produce hydrogen without CO2 generation, purifying streams of thermocatalytically decomposed natural gas. The study departed from a kinetic model obtained from differential regression of data acquired by thermogravimetry. This differential model (Avrami type) did not take into account some effects regarding the chemical equilibrium between reactants and products, neither those provided by the solid bed. To cope with this problem, some parameters were introduced in the kinetic model and experiments were performed in order to test the validity of the changes. These consisted of reduction steps with hydrogen and oxidations with steam along five alternated cycles in a fixed bed reactor. The refurbished reactor model (including kinetic model) consisted of a mono-dimensional fixed bed reactor working in non-stationary state. Initial parameter values were taken from the former kinetic model and later optimized with the aid of a Levenberg–Marquardt algorithm. The new model is able to predict with great accuracy the behaviour of the fixed bed reactor and represents an interesting tool for scale-up and process design.

► This study relates to the steam-iron process for production of pure hydrogen from reduced metals. ► Departing from a kinetic model based on empirical data obtained in a thermobalance (differential model), a reactor model has been built that takes into account chemical equilibrium between reactants and products, bed effects and loss of reactivity of the solids along cycles. ► The parameter values of the new model have been optimized by a Levenberg–Marquardt algorithm to fit the empirical results of a lab scale reactor of the same features built and run in our laboratory. ► The new model is capable of predicting the experimental evidences an can be used as a tool for up-scaling and process simulation.