Experimental And Theoretical Studies On The Mechanism Of Flashover Occurred In Long And Narrow Underground Space

Nowadays, given the contradiction between the rapid growth of population and the limited land resources in cities, many countries have focused on the development of underground space. While improving the urban land utilization, many new problems and challenges on fire safety appear. Some underground spaces are long and narrow with poor ventilation. Once it catches fire, hot smoke will spread fast along it. Besides, smoke and heat are not easy to be exhausted, so flashover, the ultimate and devastating event in a fire disaster, easily occurs in such kind of confined spaces, which may cause considerable fatalities as well as damages to infrastructure. Therefore, it is of much significance to study the fire development and the flashover mechanism in the long and narrow underground spaces (LNUS).With a series of reduced scale fire experiments, theoretical analysis and numerical simulations, the principle of fire development, the characteristic of smoke movement and the mechanism of flashover occurrence in the LNUS were studied in the present paper. At first, referring to the previous studies, a reduced scale LNUS experimental model was designed. Flashover phenomenon was reproduced in it with different fire sizes, fire shapes, fuel types, fire locations, opening sizes and ventilation modes. In these experiments, the temperature, concentration, velocity of smoke, mass loss rate of fire source as well as thermal radiation flux on the floor were accurately measured, which would benefit the validation and discussion of theatrical models and modeling codes.Based on the experimental and theatrical analysis, the thickness, the horizontal distribution of temperature and velocity of the smoke layer during pre flashover stage were discussed. On the one hand, the ceiling jet flow induced by a line fire was considered. By simplifying the three dimensional flow field into a two dimensional space, a simple model was presented, in terms of Richardson number Ri and non dimensional ceiling jet thickness h , to predict the temperature and the velocity of line fire induced ceiling jet in a LNUS. Meanwhile, the model was validated through the experimental data. On the other hand, The well founded model of unconfined ceiling jet, together with the line fire induced ceiling jet model presented in this paper can be used to estimate the reduce of temperature and velocity as well as the location of hydraulic jump along the LNUS. It is found that if the ratio of half width to height of a given LNUS is between 0.17 <λ/2 < 0.78, the critical location of hydraulic jump occurs at a distance 1.6 > xc> 0.78. According to the comparison of experimental and calculated data, the prediction of these combined models was verified well.According to the measurements of reduced scale fire tests, the smoke layer's temperature, thickness, gas concentration and heat release rate of fire source as well as thermal radiation were further discussed. Moreover, the thermal and ventilation effect to the occurrence of flashover were studies. First, it is found that the haptane pool fires represent a class of fuels that are very responsive to the thermal feedback of the LNUS, while the wooden cribs represent fuels controlled by internal combustion effects and radiant heat transfer among the sticks. Second, to some extent the heat release rate of fire source is dominated by the ventilation. If the ratio of the area of opening to the area of fire source reduces, the heat release rate will decrease.In the previous studies, the flashover phenomenon happened in ordinary compartments was primarily discussed. However, due to the discrepancy in building structures, the classical flashover model cannot be used to explain the phenomenon occurred in LNUS. Based on the conventional non linear dynamics theory, a multi section model was proposed to analyze the phenomenon. In this model, the LNUS was divided into several sections, and in each section is further divided into two zones, which are the hot smoke layer and the cold air layer. Each hot smoke layer was defined as a control volume, the conservation equations of which were then established. The temperature curve in each control volume was obtained by numerically solving the equations. Consequently, the effects of instant wall temperature and smoke layer's thickness to the section temperature were carefully analyzed, and the influences of fire radius and fuel types to the critical condition of flashover were also theoretical discussed. Finally, the theoretical critical conditions were verified by the experimental data.Finally, a validation study for a newly developed CFD code SIMTEC as well as a conventional one FDS was presented, and the modeling results are compared with experimental data which were obtained through a batch of fire tests in the reduced scale model. Besides, the prediction accuracy of three turbulent combustion models, including Mixture Fraction model,EDC model and Flamelet model, at the under ventilated condition was also compared. The simulations agree reasonably well with measured results. In general, these two codes predicted the hot layer temperature and CO₂ concentration with good accuracy, but the CO concentration was not well captured in the simulation.