Derivation of the kinetics of devolatilisation and oxidation of pulverized biomass in a drop tube furnace: Sensitivity to volume evolution and drag-coefficient model

International audienceCombustion experiments of raw and torrefied pine and demolition wood particles (600–800 µm) are performed at 800 °C in a drop tube furnace. The results provide the oxygen and carbon monoxide profiles along the reactor axis. These data are then used in a numerical model, developed to determine the kinetic parameters of devolatilisation and oxidation of the pulverized biomasses. In order to simulate the gas phase reactions, the model also takes as input the composition of the volatiles of the tested fuels measured during pyrolysis experiments at 800 °C in the drop tube furnace. The model adopts different scenarios of particle volume evolution and different drag coefficient models in order to test their influence on the derived kinetic parameters. One of the volume evolution scenarios is a specific sub-model obtained by optical diagnostics of the combustion of the three biomasses in a previous study. Four other volume sub-models found in literature are also tested. For each of these scenarios, the model estimates close activation energy for devolatilization with a maximum variation of 2 kJ·mol−1 from one scenario to another, while the activation energy of char oxidation is more influenced, varying by 14 kJ·mol−1 with different scenarios. The five scenarios show similar gas concentrations and burnout versus the distance travelled by the particle. Nevertheless, this gives rise to a noticeable difference in the particle temperature along the furnace axis (±100 °C at some positions), in addition to different particle velocity and residence time (~ ±10%). The influence of the drag force is also studied using enhanced non-spherical model versus a spherical model. The non-spherical model leads to 10 to 14 kJ·mol−1 higher devolatilisation activation energies and 10 to 19 kJ·mol−1 higher char oxidation activation energies than the spherical model, along with a better prediction of the CO levels