Approaches to Biomass Kinetic Modelling: Thermochemical Biomass Conversion Processes

Abstract

Modeling of biomass pyrolysis kinetics is an essential step towards reactors design for energy production. Determination of the activation energy, frequency factor, and order of the reaction is necessary for the design procedure. Coats and Redfern's work using the TGA data to estimate these parameters was the cornerstone for modeling. There are two significant problems with biomass modeling, the first is the determination of the kinetic triplet (Activation energy, Frequency factor, and the order of reaction), and the second is the quantitative analysis of products distribution. Methods used in modeling are either One-step or Multistep methods. The one-step techniques allow the determination of kinetic triplet but fail to predict the product distribution, whereas multistep processes indicate the product's distribution but challenging to estimate the parameters. Kissinger, Coats, and Redfern, KAS, FWO, Friedman are one-step methods that have been used to estimate the kinetic parameters. In this work, after testing more than 500 data points accessed from different literature sources for coal, oil shale, solid materials, and biomass pyrolysis using one-step global method, it was found that the activation energy generated by KAS or FWO methods are related as in the following equations: 𝐸𝐾𝐴𝑆 = 0.9629 ∗ 𝐸𝐹𝑊𝑂 + 8.85, with R² =0.9945 or 𝐸𝐹𝑊𝑂 = 1.0328 ∗ 𝐸𝐾𝐴𝑆 − 8.0969 with R2= 0.9945. The multistep kinetic models employed the Distributed Activation Energy Model (DAEM) using Gaussian distribution, which suffers from symmetry, other distributions such as Weibull, and logistic has been used. These multistep kinetic models account for parallel/series and complex, primary and secondary biomass reactions by force-fitting the activation energy values. The frequency factor is assumed constant for the whole range of activation energy. Network models have been used to account for heat and mass transfer (diffusional effects), where the one-step and multistep could not account for these limitations. Three network models are available, the Bio-CPD (Chemical Percolation Devolatilization) model, Bio-FLASHCHAIN, and the Bio-FGDVC (Functional Group Depolymerization Vaporization Crosslinking models). These models tried to predict the product distributions of the biomass pyrolysis process