Computer Modeling for Fire and Smoke Dynamics in Enclosures: A Help or a Burden?

Fire simulations are of unique value in different respects: for fire safety system design calculations, for improving our understanding (theory and model development and validation) and for fire forecasting. In this paper the effective use of computing power for the further development of fire safety science is discussed, considering only gas phase phenomena in fire and smoke dynamics in enclosures. Arguably, much effort needs to be devoted to multi-phase phenomena (pyrolysis modeling, the effect of water or other suppressants, etc.), but this is not discussed in the paper at hand. A common feature to all types of simulations is that the Required Computing Resources (RCR), determined by the envisaged accuracy and the complexity of the problem to be tackled, must be less than the Available Computing Resources (ACR). Accuracy, reliability and dimensionality of the models used, must therefore be related to the problem tackled. In order to make progress, bench-marking studies, as a joint effort made by modelers and experimentalists, with transparent communication, are argued to be a good approach for systematic progress in the development of, and confidence in, models. Using Computational Fluid Dynamics (CFD) can be very valuable for the development of theory and the study of detailed fluid mechanics phenomena, but several aspects are important to guarantee the quality of the results. Some ideas will be formulated on how to investigate requirements on the computational mesh. Turbulence - chemistry interaction (TCI) and turbulence - radiation interaction (TRI) are also discussed briefly. Yet, it is argued that a major source of uncertainty in computer simulations stems from (and will continue to stem from) the characterization of the ever changing and developing materials (i.e. the fuel), as well as from geometry dependent features (including ventilation and heat transfer). This affects the combustion and soot formation, and therefore the fire and smoke dynamics. Therefore, a user-defined fire will remain indispensible in the foreseeable future when using computer modeling for the sake of design of fire safety systems. Once this fire has been defined, CFD is best suited in regions where detail is required or complex flow patterns establish, while other forms of modeling can be considered in other regions. For real-time and forecasting applications, it is argued that sensor-assisted numerical simulations are very promising and their use is expected to become widespread in the coming decades. With increasing computing power, the use of CFD will become more feasible in this context, but for the time being zone model calculations (perhaps combined with CFD in regions where more detail is required) seem better suited to that purpose.