Influence of reaction mechanism accuracy on the chemical reactivity prediction of complex charring material in fire condition

摘要

The production of combustible gases from solids during thermal decomposition and how these gases feed the flame take an important part in the currently fire research. In this context, the modelling of the thermal, chemical and physical phenomena in the condensed phase constitutes a key step to accurately predict the gases production (source term). This modelling applies to establish a relationship between the kinetics of gases production (thermal decomposition of the matter), the solid temperature and the atmosphere composition (% of oxygen). Due to digital progress, the accuracy of modelistic approaches does not depend on the resolution technique, but rather on the accurate description of the apparent chemical reaction pathway. This mechanism, identified through the matter scale experiment, will define the number of kinetic parameters to be estimated. This study focuses on the complexity level of the reaction mechanism, and its effect on the accuracy of the chemical reactivity forecast. The analysis is thereby performed on a solid material (polyisocyanurate foam) with a complex multi-step mechanism. Several mechanisms complexity with various levels of sophistication (number of linked reactions) are studied while their results are presents and discussed. This work uses a new statistical technic based on statistical criteria to judge the quality of the model and to choose a reaction mechanism with a degree of complexity adapted to fire modelling at large scale. The analysis done reveals a possibility to simplify the initial reaction mechanism's complexity, without influencing the accuracy of forecasts. These works have therefore demonstrated that the use of highly simplified mechanisms enables the reproduction of the complex decomposition kinetics of the PIR foam. On the other hand, the use of a detailed reaction pathway is not always necessary to predict the decomposition of complex material through modelistic methods. (C) 2016 Elsevier B.V. All rights reserved.