Comparative analysis of pyrolysis models including SFOR, CRECK, and Bio-CPD to predict reaction kinetics and products from extracted biomass components

Biomass pyrolysis is typically modeled based on the three reference components cellulose, hemicellulose, and lignin. Most models rely on an individual decomposition of the materials and a linear superposition of the individual component products weighted by the present mass fractions. Models of varying complexity exist for the mathematical description of the pyrolysis process, ranging from the simplest single first-order (SFOR) model and the multi-step CRECK model to the chemical percolation devolatilization (CPD) model representing the molecular network of the solid. The models differ not only in their complexity but also in the used data for initial parameter calibration — thermogravimetric analysis (TGA) data for the CRECK model and mainly entrained flow, fluidized bed, or wire mesh reactor data for the Bio-CPD model. Within the present study, the predictive performance of these three models is compared with regard to the time-dependent total volatile release and the final volatile yield when applying two different thermal boundary conditions: low heating rate in a TGA and flash pyrolysis conditions realized with a small-scale fluidized bed reactor (FBR). The models are compared with one other and with experimental data on extracted, separately pyrolyzed biomass components to examine under which conditions reliable predictions can be made and when the trustability is limited. For the TGA data, the CRECK model has the closest proximity to the experimental measurements, also resolving most of the individual reactions, while the SFOR model can resolve only one globally dominating reaction, and the Bio-CPD model strongly overpredicts the reactivity of the biomass components during the slow heat-up. Under flash pyrolysis conditions in the FBR, by contrast, the Bio-CPD model predictions are closest to the experimental results when it comes to predicting the total volatile release rate. However, examining the integrally released yields, the CRECK model is closer to the experiments. Regarding the tar and light gas distribution, all models strongly overpredict tar from primary pyrolysis compared to the experimental results, indicating the presence of secondary gas-phase reactions in the FBR. Although different secondary gas-phase reaction models are used, the tar yield is significantly overestimated by the models compared to the experimental data.