Application of a simplified pyrolysis model to predict fire development in rack storage facilities

Abstract

The simplified (thermal) pyrolysis model is applied to simulate flame spread and fire growth in the rack storage facility using the FDS software. Pyrolysis in the combustible material is not explicitly considered, and the ignition temperature and the burning rate are used as the input parameters. Once the fuel surface temperature reaches the ignition temperature, the material ignites and burns at a prescribed burning rate. This approach is found to reasonably replicate both the transient development of the heat release rate and flame dynamics observed in the full-scale rack storage fire with 2x4x3 cardboard boxes in the rack, provided the model parameters (thermal properties of the fuel, ignition temperature, average heat release rate per unit area, burn-out time, and heat of gasification) are properly selected.Distinct flame propagation regimes including buoyancy-driven upward flame spread, horizontal flame spread, and buoyancy-opposed downward flame propagation are observed in the simulations of the rack-storage fire development. Both measured and predicted HRR growth rates appear to be faster than that in a t-squared fire, mainly because of the developed combustible surface densely packed within the compact volume and due to availability of vertical gaps creating chimney-like effect and horizontal gaps providing permanent supply of fresh air.It is shown how the full-scale test data enable selection of the model parameters. In particular, average value of the heat release rate per unit area is evaluated by joint consideration of the measured dynamics of the growing heat release rate and the surface area engulfed in fire. The burn-out time and the effective thickness of the fuel layer is estimated as the period from ignition to the time instant after which the measured heat release rate starts to decay. The heat of gasification is selected by fitting the predicted transient dependence of the heat release rate to that measured in the full-scale tests. Optimum values of the model parameters are consistent with the literature data for cardboard, and this indicates possibility of the simplified thermal pyrolysis model to predict rack storage fire dynamics with an alternative fire load, for which the full-scale test data may not be available. Further work is required to validate this approach for a wider range of full-scale fire tests.