Coupled hybrid modelling for fire engineering
Ralph, Benjamin Michael
Due to time and cost constraints, fire engineers typically curtail the domain of analysis when carrying out quantitative assessments using computational fluid dynamics (CFD)-based fire modelling. This could embody unquantified hazard and is especially critical when designing complex building with a shared ventilation system. Secondarily, prescriptive model solutions have been developed using engineering judgement and empirical evidence. There is a risk that under-investigated mechanisms may lead to unacceptable prescriptive solution risk levels. The overarching thesis aim is to enable a more robust quantification of fire hazard in complex buildings with a shared ventilation system. The objectives of the study are to develop and evaluate a novel coupled hybrid model implementation and highlight potential shortcomings of existing prescriptive design solutions for shared ventilation systems. This study develops a novel 3D-1D coupled hybrid modelling implementation within Fire Dynamics Simulator (FDS). The new implementation addresses the pre-existing limitation of time-dependent transport or storage within the 1D sub-domain and introduces a novel fan model. A new experimental rig was created comprising two 1m3 boxes interconnected with shared mechanical ventilation. A propane burner was used as a fire source, with propane flow rates ranging between 0:2 g=s to 0:45 g=s, in one of the boxes. Variable speed controllers and dampers were used to alter the ventilation with target free flow fan velocities of 1m=s to 3m=s. The novel model implementation satisfactorily passed verification and presented generally good agreement with the experimental results. Prediction of maximum temperature in the fire and non-fire enclosures are typically within 40% and 5% respectively. Prediction of ventilation duct velocity is typically within 5 - 25%. Model correction factors of 1.0, 0.7, and 1.4 are proposed for enclosure temperature, in-duct temperatures, and duct velocities respectively. Experimental data demonstrate that empirical methods may not be suitable for complex arrangements with shared ventilation because there is a stronger dependency of fire hazard upon the ratio of heat losses to ventilation enthalpy advection when compared to traditional arrangements. The data illustrate that remote area fire hazard is very sensitive to the balance of the energy transfer rate (i.e. power) of the fire and the ventilation system and that this relationship is non-linear; a correlation which would not be well-captured using the typical modelling paradigm nor prescriptive design solutions. The study concludes that the new coupled hybrid modelling implementation may be used to analyse a total system with quantified uncertainties. Further development is recommended for the new model implementation (e.g. conductive heat transfer within the 1D sub-domain). Further experimentation is recommended to further inform prescriptive design solutions for complex buildings with a shared ventilation system.