System Identification of SOFC Oxygen Reaction Parameters by Distributions of Relaxation Times

Abstract

System identification, a method used in control theory for modelling complex dynamical systems, is applied to the identification of SOFC oxygen reaction parameters. The present approach is based on distributions of relaxation times, which are the basic quantity of interest in electrochemical impedance data analysis. By a newly implemented deconvolution method, characteristic distribution patterns are directly computed from impedance spectra measured at single cells under realistic working conditions. In contrast to non-linear least squares curve fit of equivalent circuit models, no a priori circuit choice has to be made. The excellent resolving capacity allows to untangle impedance contributions of up to three physically distinct processes within one frequency decade. The analysis procedure is illustrated by a reaction model which describes the adsorption of oxygen on the electrode surface and the incorporation of oxide ions into the electrolyte. Based on the reaction model, the influence of varying the cell's operating conditions on the parameters of the relevant peaks in the distributions is simulated. The kinetics of the reaction steps are then identified using distributions of relaxation times calculated from the impedance spectra measured at the corresponding operating conditions, thus providing a strong tool for SOFC diagnosis.