A critical comparison between CFD and zone models for the consequence analysis of fires in congested environments

Fire represents a very complex phenomenon, as the severity of the consequences associated with its evolution depends not only on the fire source itself, but also on the configuration of the system in which it develops, that is whether the source term is placed in an unconfined environment or it is confined in a room or in an elongated space such as a tunnel, either empty or filled with vehicles. Numerical modeling approaches allow to obtain cost-effective and reproducible information, with a degree of confidence dependent on the quality of the input information and the assumptions made by the model itself. In this study two different modeling approaches have been compared, a computational fluid dynamics code and a zone model in order to test their effectiveness and their limitations in predicting different congested fire scenarios. A series of tests has been performed using a single-room configuration that is the most classical application of zone models, comparing the models performances with experimental results and highlighting the scarce sensitivity of zone models to some aspects, such as the source term position. A simple tunnel test has then been reproduced, testing the different models capabilities of identifying general trends and fundamental parameters, such as the critical ventilation velocity and the temperature profiles along the elongated geometry. Moreover, both models have been used to predict the effect of the presence of obstacles in the computational domain, showing that zone model cannot account for their presence. The simulation results highlight that, although the simulation times and the expertise required in order to implement and process the information are much lower, zone models can be useful only in very simple configurations; moreover, they can be extended with some limitation to complex room fires and cannot be relied upon for tunnel fires analysis, where fundamental aspect such as the critical ventilation velocities are not correctly predicted.