Modeling of the pyrolysis of plywood exposed to heat fluxes under cone calorimeter

In this paper, the thermal decomposition of plywood is investigated based on the solid mass loss rate (MLR) modeling. The multi-scale approach followed here allows first to establish, at a small scale, the kinetic mechanism during the solid thermal decomposition and then validate it at a larger scale. At small scale, experiments were conducted by using Thermo-gravimetric analysis (TGA) coupled to gas analysis with the FTIR technique under nitrogen and air atmospheres for five heating rates. Thermo-gravimetric results were also used to propose a kinetic mechanism for the thermal decomposition of the sample. The kinetic parameters of the different identified reactions were estimated by using an optimization technique, namely the Genetic Algorithms (GA) method. The mass loss rate model predictions show a good agreement with the experimental data. At a larger scale, experiments were carried out in a cone calorimeter coupled to FTIR gas analysis under air atmosphere. The pyrolysis model developed at the TGA scale was used in numerical simulations of cone calorimeter experiments taking into account the heat transfer modeling into the sample. The thermal properties (e.g. thermal conductivity and specific heat capacity) of the condensed phase species (which are products given by the thermal decomposition of the virgin material) identified at the TGA scale were estimated for an incident heat flux of 30 kW.m -2 with the same optimization technique used at the small scale: the GA method. The same work has also been conducted with a simpler well known pyrolysis model developed for charring materials. The results have been compared to those of the first mechanism in order to show the influence of the complexity of the model on the prediction of the thermal decomposition of plywood. The heat transfer model was kept the same for both pyrolysis models. Only the number of identified condensed phase reactions and species is different. The detailed mechanism (5 steps) gives better results than the simpler one (3steps) concerning the mass loss rate prediction and worse results for the temperature prediction of the back surface of the sample. However the 3 steps model gives unrealistic results concerning the prediction of the condensed phase species mass fractions. In fact, the 3 steps model predicts that plywood is not fully burned at the end of the test (for 30 kW.m -2 ) which was not observed in the experiment.