Model‐based analysis of chemical‐looping combustion experiments. Part I: Structural identifiability of kinetic models for NiO reduction

Abstract

To guide the design of chemical-looping combustion (CLC) systems, the use of accurate models is crucial. The reduction kinetics between NiO and CH4 is uncertain, in regards to the most suitable kinetic mechanism and reaction network. A framework for structural identifiability analysis is developed and applied to evaluate the candidate kinetic models for the NiO-CH4 reaction. The identifiability of kinetic parameters of different model structures is analyzed and compared. Models that lack structural identifiability of their kinetic parameters are rejected in the analysis. From a total of 160 possible candidate models, 4 kinetic models are found to be identifiable with respect to their kinetic parameters and distinguishable from different model structures. This structural identifiability analysis paves the way for model-based design of experiments, which is the subject of Part II of this work.

Model‐based analysis of chemical‐looping combustion experiments. Part II: Optimal design of CH4‐NiO reduction experiments

“…A1, were previously estimated from the nominal reduction experiment presented in Figure 4 of Part I. 24 The activation energies for R2-R4 were fixed to the reported values of Zhou et al 1 and for R5 to the reported value of Dueso et al 9 For the catalytic reactions, k Tref i was estimated in Part I, 24 keeping Ea i fixed to the reported values of Zhou et al 1 The results of the optimal experimental designs for model discrimination and parameter estimation are discussed. The posterior statistics of the parameter estimates are presented for each step of the framework.…”

“…6, in which all models deemed identifiable in Part I 24 were assumed equally likely to be the true model. Equation 6 was solved over the four candidate models with bounds on the parameter values h Table 3.…”

“…The kinetics of the catalytic reactions were determined from the nominal low-temperature experiments studied in Part I. 24 Methane decomposition reactions and the impact of carbon formation on the catalytic reactions are possibly not perfectly captured in the corresponding kinetic parameters, due to the excess of steam in the reforming experiments. Thus, in the D-optimal predictions of Figure 11 the CH 4 fraction was slightly over-predicted.…”

“…In total, 160 possible kinetic models exist for the reduction of NiO by CH 4 . In Part I 24 of this work, we developed and applied a systematic framework to analyze the structural identifiability and model distinguishability of the candidate kinetic models. We determined that the simplest catalytic reforming network (RS IV) and the linear model for the catalytic activity of Ni (Eq.…”

Dynamic Simulation of Fixed‐Bed Chemical‐Looping Combustion Reactors Integrated in Combined Cycle Power Plants

“…The Ergun equation was used for the momentum balance. The reaction kinetics was derived from atmospheric‐ and high‐pressure gaseous CLC experiments that used supported Ni‐ and Cu‐based oxygen carriers . This model was developed by using gPROMS and was used successfully for the prediction of CLC data over a range of operating pressures, temperatures, and fuel compositions.…”