Bayesian Design of Experiments for Adsorption Isotherm Modeling

Abstract In this study, a Bayesian approach is demonstrated to identify parameters in adsorption isotherm models. First, we demonstrate the data fusion with Bayesian analysis for which we use CO2 sorption on UiO-66 sorbents. This material has been studied by different groups to report CO2 adsorption data in the NIST database under different conditions. We unify the data in two different ways depending upon the assumption around the error distribution of the data. In the first case study, all the data are assumed to have a common error distribution regardless of the source, and in the second approach, the error distributions of data from different sources are distinguished from each other. We apply a Bayesian approach to perform an optimal experimental design to minimize the uncertainty in adsorption isotherm parameters.