A Comparative Analysis of the Use of Different Zone Models to Predict the Mass Smoke Flow for Axisymetric and Spill Plumes

Peer-reviewed article published in the Proceedings of the 9th International Symposium on Fire Safety Science, Karlsruhe, 2008.Sanderson (2007) examined the theoretical basis for and the experimentation supporting the predictive smoke zone models currently being used in fire engineering design that are cited, in nationally and internationally accepted guidance documents, to support the increasing use of performance-based building codes/regulations throughout the world. This critical examination identified anomalies: 1) between different researcher’s results, when considering the same fire environment, and 2) areas where the models used in guidance documents have limited empirical support. The variance between models was examined by the parametric variation of critical data input parameters the impact of which indicated that the most recent research produced models that predict a lower level of mass smoke flow than the earlier models. It may be suggested that the more recent research, building on previous work, produces models that can be used with a greater level of confidence however there is no robust evidence to support this. This paper illustrates the variances between the model outputs by means of a case study. Currently there is a move towards the use of Computational Fluid Dynamic modelling of fire. However, given the limited validation of these models in the area of smoke movement and the computer time and power required to run these models, there is still a place in fire engineering design for the zone model. As an increasing number of countries adopt performance building and fire codes/regulations and given the consequent need for predictive mass smoke flow models in which regulators, fire engineers and society can have confidence, it is concluded that the research supporting zone modelling of fire should be extended. This research should be robust and transparent in order to either produce models that are substantially more acceptable than those currently being used or to provide more confidence in models