A model-based experimental design approach to assess the identifiability of kinetic models of hydroxymethylfurfural hydrogenation in batch reaction systems

Abstract

Hydroxymethylfurfural (HMF) is an organic compound that occurs naturally in many foods and is used as feedstock in numerous chemical processes. HMF can be hydrogenated to form 2,5-Dimethylfuran (DMF), which is an important component in biofuel production. To date, several kinetic models have been proposed and studied in literature for this hydrogenation reaction, including power law models based on reaction species and Langmuir-Hinshelwood-Hougen-Watson (LHHW) models. For these models a critical aspect that has not been addressed in literature is related to their practical identifiability, i.e. the estimability of kinetic parameters from experimental data. Also, none of the existing models propose a temperature dependence of the kinetic parameters.A three-step approach is presented in this paper, which exploits model-based design of experiments (MBDoE) techniques to assess the identifiability of candidate kinetic models of HMF hydrogenation in a batch reaction system. The objective is twofold: (1) to propose new kinetic models of HMF hydrogenation where the temperature is explicitly introduced as an experimental design variable and test the practical estimability of kinetic parameters from concentration data only; (2) to identify the most informative regions of the experimental design space, defined by temperature, experiment duration and initial HMF and DMF concentrations, for achieving a precise estimation of model parameters. Together with a-posteriori statistics obtained from parameter estimation from in-silico data, an MBDoE analysis gives a clear representation of the most informative experimental conditions to be explored in the future experimentation underlining distinct areas of practical parametric identifiability.