Travelling fires methodology and probabilistic design of structures

Abstract

Fire is a hazard and a building’s structure must be designed to maintain their structural stability when exposed to fire. Fire safety design of structures can be done following the prescriptive codes or carrying out a performance-based design. Prescriptive fire design codes describe how buildings should be built to fulfil generic fire resistance requirements depending on their use, height, or compartment area. Performance fire design allows derivation of structural fire resistance of buildings by characterizing the fire dynamics within a compartment and analysing the thermal and mechanical response of the structures. Travelling fires methodology (TFM) characterises the fire dynamics for the performance-based design of large compartments, which assumes that as fires burn, they travel along the compartment floor as flames spread. TFM is a design tool based on several assumptions. This thesis revisits and addresses its near field assumptions and applies a probabilistic model to assess the reliability of a structural element exposed to travelling fires and the uniform temperature fire. This work studies the horizontal flame extension under the ceiling, which affects duration of the heating exposure of the structural members and their load-bearing capacity. This study reformulates the TFM in terms of heat fluxes rather than temperatures, allowing for a more formal treatment of heat transfer. The Hasemi, Wakamatsu and Lattimer empirical expressions of heat flux from flames were applied for the near field. The analysis showed that the near field length with flame extension (fTFM) is between 1.5 and 6.5 times longer than without flame extension meaning that more structural elements are affected by the direct impact of the flame. The highest peak heat flux is obtained for small fires sizes, using flame extension and the Wakamatsu expression. The observations and findings from two travelling fire experiments, x-TWO.1 and x-TWO.2, conducted inside a very large compartment with an area of 380 m² in Poland, are presented. A uniform continuous wood crib along 29 m of the 35 m compartment provided the fuel load density of 345 MJ/m² in x-TWO.1 and 273 MJ/m² in x-TWO.2. In both experiments, the fire was observed to travel with clear leading and trailing edges and flashover was not observed. Flame spread was accelerating in x-TWO.1 and at a constant rate in x-TWO.2. The non-uniform distributions of temperatures are remarkably different from the conditions typically assumed in other scenarios/standards and could therefore lead to different failure times and mechanisms. A detailed sensitivity study was carried out for the main input parameters in the uniform temperature fire methodology, and in the heat transfer calculations. This was done to assess the sensitivity of structural element response to input parameter uncertainty. The sensitivity analysis identified the most influential input parameters for the structure and the range of values for each input parameter for which the design is structurally safe. To study comprehensively the structural fire design of open plan compartments and modern buildings, a simple probabilistic methodology was applied. Probabilistic approach defined the reliability of a fire-affected structure following the uniform fire condition methodology and fTFM methodology, while accounting for uncertainties in input parameters. The probabilistic analysis allowed quantification of the reliability of a structural element in terms of the likelihood of collapse for different fire scenarios. This study demonstrates that both sensitivity and probabilistic analyses can provide a comprehensive understanding of the factors affecting the structural fire resistance of a building and can further inform detailed fire safety and structural analysis. This work presented in this thesis is dealing with design purposes of non-combustible structures.