Since fires can develop from endogenous smouldering combustion processes deep inside packed beds, especially domestic refuse beds, it is a major task in early fire detection to detect the indications of such combustions as early as possible. Since it is believed that the surface temperature distribution of the bed is affected by the heat and mass transfer from a source of endogenous combustion deep within the bed, the measurement of the surface temperature using IR-Thermography (IRT) has been supposed to be the most promising technique in early combustion detection. The present work thus deals with the heat and mass transfer from endogenous sources of combustion in packed beds, particularly domestic refuse beds, in order to predict the temperature distribution inside and at the surface of these beds and thus, to allow for an assessment of IRT based early combustion detection systems. An IR-thermographically measurable surface temperature increase will only be achieved by sufficient heat transfer from the source of combustion. Experimental procedures and mathematical modelling have shown that the heat transfer by conduction and radiation is ineffective and therefore, no indication will be obtainable from the surface temperature distribution. A more satisfactory increase in the surface temperature is given as soon as additional heat is transferred by the diffusion of gaseous combustion products. As a result, the heat transfer by convection from the hot combustion gases is theoretically analysed, in particular the way in which the gases flow from the combustion to the surface of the bed. The results obtained show that the gas flow is initiated and maintained by buoyancy and thus the gas tends to flow vertically towards the surface with minimal collateral diffusion. It was also shown that heterogeneous polydispersed beds can be treated as homogeneous monodispersed beds as long as average values for the characteristic bed properties can be obtained. Based upon that, a mathematical continuum model was derived by which the temperature distribution inside and at the surface of the bed could be predicted. The theoretically obtained results and their implications were then experimentally verified, confirming that heat is predominantly vertically transferred by the gaseous combustion products. The two final sets of experiments were undertaken in a 27 m3 batch of a representative sample of domestic refuse and a batch of wood chips. Comparing the experimental results with the mathematical predictions, a certain deviation becomes apparent, which is attributed to the not exactly one-dimensional condition inside the bed and especially to the effect of condensation and re-evaporation of the water content of the combustion gas, which has not been included in the model. The temperature of hot spots at the surface, that have the size of only a few square-centimetres, increased to the dew point temperature of the combustion gas, i.e. 65 °C to 85 °C, within the first hour after ignition but remained almost constant at this level for several hours. About 30 minutes before the combustion proceeded to the surface, their temperature increased rapidly but not their size. The rapid temperature increase was attributed to the condensation and re-evaporation ceasing because the entire bed had heated up to a temperature at which these effects no longer occur. All results obtained, especially for the surface temperature development, allow for an assessment of IRT based early combustion detection systems. Whilst, on one hand, sources of endogenous combustion are principally detectable from the surface temperature distribution, on the other hand, the reliable detection of the very small hot spots requires a spatial resolution of the system, which is up to ten times higher than recent state-of-the-art systems can provide.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This study compared experimental and Computational Fluid Dynamics (CFD) results for heating and cooling in a packed bed (N=5.33). The experimental data was compared between heating and cooling, and was also used to validate the CFD model. The validated models were used to compare theoretical heat transfer parameters. For the experiments, it was found that the effective thermal conductivity was comparable for heating and cooling, and the wall Nusselt number for heating was higher. For the CFD results, it was found that both the wall Nusselt number and effective thermal conductivity were comparable for heating and cooling. The wall Nusselt number was slightly higher for cooling, however this difference decreased as the Reynolds number increased.

This study compared experimental and Computational Fluid Dynamics (CFD) results for heating and cooling in a packed bed (N=5.33). The experimental data was compared between heating and cooling, and was also used to validate the CFD model. The validated models were used to compare theoretical heat transfer parameters. For the experiments, it was found that the effective thermal conductivity was comparable for heating and cooling, and the wall Nusselt number for heating was higher. For the CFD results, it was found that both the wall Nusselt number and effective thermal conductivity were comparable for heating and cooling. The wall Nusselt number was slightly higher for cooling, however this difference decreased as the Reynolds number increased.