Simulation of the pyrolysis of charring polymers: influence of the porous media properties
Résumé
Recent large-scale experiments of cable tray fires have shown that the composition of the cable-sheath materials have a strong influence on the fire growth and spread. Thus, the correct prediction of the fire scenario requires the characterization of the sheath materials thermal and thermokinetic parameters. Yet, this task may be tricky for the materials whose pyrolysis generally leads to the formation of a porous residual. Whereas the physical constants related to the material degradation kinetics and to the virgin material properties can usually be determined using thermogravimetric analysis and differential scanning calorimetry, the porous residue properties are difficult to obtain and rarely available for pyrolysis models validation.
The present paper deals with the simulation of reference cone calorimeter experiments carried out on PVC samples. An extension to porous media of the pyrolysis model of the CALIF3S-ISIS fire simulation software, developed at IRSN and validated on non-charring polymers, is used for the simulations. Special attention is paid to the modelling of the thermal diffusion effects in the porous residuals. On the one hand, based on the intrinsic conductivity of the different media (virgin material, intumescent intermediate state, char and gases), several effective conductivity models are tested to evaluate the relevant range for the global material conductivity. On the other hand, it is observed that the porous residuals of PVC generally contain large pores due to the material intumescence. That is why a radiative conductivity is included in the model to account for the radiative transfer in the pores. Without precise knowledge of the PVC residuals morphology, a Penetrable Sphere Model (PSM) is chosen to represent the porous medium geometry. This model is parametrized by a constant porosity deduced from the experimental data, and a variable pore size. It allows deducing a porous medium radiative conductivity which is added to the usual effective conductivity.
It is shown that, in the particular case of PVC pyrolysis, the uncertainty on the purely conductive effects induces a very limited dispersion on the simulation of the Heat Release Rate (HRR), since all the considered models yield very similar results. Conversely, the sensitivity of the radiative conductivity and therefore of the pyrolysis kinetics to the PSM pore size is much larger. The optimal pore size with respect to the HRR temporal evolution is of the same order as the pore size observed in some PVC residuals. Finally, pyrolysis simulations are performed with this optimized parameter for different cone heater induced heat fluxes and sample sizes. A good overall agreement is found between the experimental and the numerical HRR evolution, except for the thinnest sample for which the numerical HRR is overestimated. More generally, this work shows that further effort on the morphological description of the porous residuals and on the modelling of the radiative transfer is required if one aims at performing predictive pyrolysis simulations.