INVESTIGATION OF THE REDOX PROPERTIES OF COMMERCIAL NICKEL CATALYSTS AND ITS CONSEQUENCES FOR COMMERCIAL SYNGAS PRODUCTION PROCESSES

Abstract

A thermodynamic model was developed in order to analyze the effect that different process schemes for synthesis gas production and for nickel catalysts reduction procedures have on the catalyst. This model is a first approach in the development of a process-design tool. It was implemented in MATLAB based on the Gibbs energy minimization method and calculates the equilibrium state between gas and solid (catalyst) phases in an isothermal reactor, giving the final nickel oxidation degree (percentage of nickel in the oxidized form) and the amount of coke at equilibrium, both parameters related to the catalytic activity. Diffusion and kinetic effects on the catalyst are not considered in this stage of the model. The model takes as input the temperature, pressure, and initial solid composition in the reactor and the composition of the gas that enters into the reactor. The fugacity coefficients for the gas phase are calculated through the Soave-Redlich-Kwong equation of state, whereas the activity coefficients for the solid phase are calculated using the Margules activity coefficient model because of its simple implementation. Before using the tool it was necessary to fit two parameters: the activity coefficients at infinite dilution of nickel in carbon and nickel in oxygen. Both parameters were fit using experimental nickel oxidation degree data from experiments carried out at the Air Liquide’s Research Center in Frankfurt using a least square minimization problem. Model results seem not to be too far of the reality. However, it is rather recommended not to rely on absolute numbers but rather on the trends being predicted by the tool. For the best cases (outlet HT-Pre-reforming and inlet SNG) the model predicted a nickel oxidation degree of 14 and 13 NiO mol%, while the corresponding experimental results were 14 and 12 NiO mol%, respectively. For the worst case (outlet SNG) the tool predicted a 0.3 NiO mol%, while the experimental result showed 14 NiO mol%. This error could be explained on one hand by an under-estimated steaming (oxidation) rate during shutdown. On the other hand model simplifications as the non-consideration of diffusion, particle size, or the particle surface free energy can also lead to wrong results. The tool succeeded to perform equilibrium calculations for catalysts and gases in (pre)-reforming and SNG processes. Now it sets a basis in the development of a more accurate tool, which can be based on the actual code since it was written to give flexibility for further changes: it allows changing functions in the code, e.g. the activity coefficient calculation.