Pyrolysis Model Development to Predict Ignition Behavior of Thermal Insulation Materials

Wildland-urban interface homes are bombarded with embers during proximate wildfires. Although ember intrusion into ventilated spaces is common, little is understood about the mechanisms of ember penetration and subsequent ignition of combustibles in attic spaces. Improvement in building codes and standards would benefit from comprehensive pyrolysis models that can accurately predict cone calorimeter as well as ember ignition behavior of thermal insulation materials.  In this work, an approach is developed to generate pyrolysis models for nine commercially available insulation materials: fiberglass, kraft-faced fiberglass, rockwool, denim, cellulose, extruded polystyrene, expanded polystyrene, polyurethane, and polyisocyanurate.  As a first step towards the necessarily three dimensional ember ignition processes, we develop a one dimensional model for uniformly heated surfaces, which is the case in cone calorimetry. The pyrolysis code GPYRO is employed to provide the volatile mass flux from the irradiated samples. GPYRO is robust in that it can handle both thermoplastics and charring solids as well as heterogeneous and homogeneous reactions. We assess the predictive accuracy of the GPYRO results against cone calorimeter results for these insulation materials. GPYRO can account for intra-particle diffusion of gaseous reactions, chemical reaction, and intra-particle diffusion of gaseous products, which allows for better agreement with diffusion controlled reactions (common at high heating rates). Thermophysical properties of each material have been determined through direct measurement and literature, optical properties were estimated through literature, and kinetic parameters and heats were estimated with simultaneous thermogravimetry analysis and differential scanning calorimetry (TGA/DSC) (3 mg, 20 K/min, oxidizing atmosphere). The decomposition chemical reaction rate for a discretized unit is determined based on the temperature, mass fraction, and gaseous reactant(s) availability (if applicable). The reaction rate follows the Arrhenius function coupled with an nth order or diffusion reaction model. Numerical optimization via shuffled complex evolution (SCE) method is applied to best fit cone calorimeter mass loss rate results at 15, 25, 50, and 65 kW/m². It is shown that a two-step, first order reaction is sufficient (within experimental uncertainty) to describe most materials. In future work, these optimization derived pyrolysis models will be applied to simulate ember ignition of insulation with comparison to empirical data.