A review on prediction models for full‐scale fire behaviour of building products

This study aims to give an overview over different reaction-to-fire prediction models developed over the last decades by finding similarities and differences between models, as well as identifying their robustness in scaling. The models have been divided into four categories - empirical, thermal, polynomial and comprehensive - depending on how pyrolysis is modelled. Empirical models extrapolate bench-scale test results to larger scales. These models are pertinent to applications that they have been validated for, but surfacic parameters used may not be scalable. In thermal models, pyrolysis is represented by heat transfer rates. The models are feasible for materials with high activation energies and where little pyrolysis occur before ignition. Polynomial models are empirical models that also take the environment into account. The validity of scaling is yet to be established. The comprehensive methodology includes chemical kinetics in the condensed phase. It has the potential to be used for any application; however, many parameters are needed. This increases the degrees of freedom versus data available for the description of the problem. Consequently, possible errors are introduced, and uncertainty is increased. A comprehensive multi-scale methodology is a way forward, where many steps of validation are possible. (Less)