It is well known that fires of different fuel type result in quite different proportions of heat loss to convection a nd radiation. For instance, the radiative loss fraction from a well-ventilated fire has been measured as 20% for methanol and 35% for kerosine [Babrauskas (1988)]. For accuracy of prediction, it is vital to properly represent such variation in the CFD models, for example in the simulation of design fires for 'fire safety engineering' methodologies.
One method of doing this is simply to reduce the fire size (heat release rate) by an appropriate radiative loss fraction. This empirical approach is routinely adopted within the (in-house) JASMINE CFD code. However, use of this method means that the overall mass flow rates and radiative feedback to the fuel surface and the enclosure boundaries are not correctly reproduced, thus introducing other modelling inaccuracies. More robust and realistic predictive methods depend on achieving an accurate representation of the flowfield soot concentrations and consequent radiative loss. Though radiation sub-models of this type are available in the SOFIE CFD code, which was developed by the consortium of fire research organisations for the fire research community, these were developed in the context of bench-scale tests and their general applicability to full-scale fire problems is yet to be demonstrated. The capabilities of the main commercially-available CFD codes in this area are similarly unproven and very limited in terms of number of fuel types which can be dealt with.
Objectives and Methodology
The objectives of the proposed study are as follows:
The treatment of the luminous and non-luminous radiation inside the fire plume within the CFD codes JASMINE and SOFIE will be revisited. By using the mixed-grey gas representation of the luminous and non-luminous radiation, JASMINE will be developed further to incorporate a "prescribed scalar" representation of flow-field soot. The radiation model will be modified to correctly represent the enhanced radiative loss due to the soot radiation. By contrast, the CFD code SOFIE already offers a more advanced flamelet-based approach for the combustion, soot and radiation processes, and conjugate heat transfer to boundary walls. However, the generality and reliability of the coupled soot and radiation model is not well-established for full-scale fire scenarios.
In order to establish a baseline on predictive capabilities, each model representation will be tested and evaluated in simple well-ventilated full-scale fire scenarios for which there is reliable experimental data on the breakdown of the overall heat loss. Since JASMINE and SOFIE differ further in their numerical formulation of the partial differential equations controlling the fire and smoke spread, the relative performance of the two codes, based on mixed-grey gas radiation representation, will also be evaluated.
Finally, the recent literature on the representation of soot formation in flamelet-linked models, in particular the effect of temperature fluctuations, will be reviewed, e.g. Brookes (1999). The predictions possible on the basis of such alternative modelling representations will be carefully evaluated so as to establish whether any improvement can be achieved for the case of full-scale fires. Where feasible, further model development would be undertaken to incorporate any necessary improvements.
Recommendations on appropriate methodologies will be communicated to the fire engineering community.
Back to top
They were initially set up by S.Welch at BRE, October 2000. Copyright (c) BRE 2000