INVESTIGATIONS ON SOLID FUEL FIRE IN A COMPARTMENT

Abstract

The present research work details an experimental investigation to enhance the understanding of the burning behavior of cribs under different ventilation conditions and crib patterns. Several full-scale crib fire experiments were carried out in a compartment having internal dimensions of 4.0 m length by 4.0 m width by 4.0 m height. A door opening of 1 m width by 2 m height was provided for the ventilation. Several measurements were made such as the mass loss rate, the centerline flame temperature, the total heat flux, the compartment gas temperature, the doorway temperature and the concentration of CO, CO2 and O2. In addition, the burning characteristics of crib fire were explored under the hood. The impact of sudden ventilation on the burning behavior of cribs was investigated. The experimental results indicated that the sudden ventilation has a significant impact on the mass loss rate of the crib. As the time of sudden ventilation provided to the compartment increases, the peak mass loss rate decreases due to decrease in the fuel load. The sudden ventilation provided to the compartment results in the rapid increase in the gas temperature at the upper zone, whereas the temperature decreases rapidly at the lower zone due to the entrainment of fresh air. Similarly, the intensity of the total heat flux increases within the compartment boundaries except the lower zone which tends to decrease due to decrease of the thickness of hot gases. A series of fire experiments was performed using plywood cribs as fire sources in a compartment. In particular, the total of four experiments was conducted. Out of them three experiments were performed for the burning of regular crib patterns of 0 %, 10 % and 100 % compartment door openings. The fourth experiment (Test 8) was conducted for the staggered crib pattern with 100 % door opening. The rate of increase of heat release rate (HRR) during growth period for staggered pattern crib (Test 8) was 11.5 kW/min, whereas for regular pattern crib (Test 7) it was 8 kW/min. However, the peak HRR was only 15 kW higher as compared with the regular pattern crib. Increasing the ventilation to 10 % and 100 % has resulted in increase in the HRR by 61.6 % and 77.4 % respectively. The compartment gas temperature and the total heat flux were highest for 10 % door opening, whereas the no significant changes were observed in flame temperatures with different ventilation conditions as well as crib pattern. Assessment of HRR in a compartment fires is a major concern for fire investigators to predict the fire behavior. Using the fundamental energy balance equation an attempt has been ii made to determine the HRR for various intervals of time. Analyzing the results it was determined that major portion of the energy release during fire went to heat the compartment boundaries followed by energy loss through compartment door opening and energy went to heat up the gases in the compartment. Nearly 62 % of the energy went to heat compartment boundaries, 27 % of the energy loss due to outflow of hot gases through door opening and 11 % of energy went to heat the compartment gases. Three-dimensional numerical simulations were performed using large eddy simulation with different values of Smagorinsky sub-grid scale constant. In order to precisely investigate the behavior of fire within the compartment by computational fluid dynamics (CFD), it is vital to verify these Smagorinsky sub-grid scale constant. Numerically predicted values with various Smagorinsky sub-grid scale constants were compared with the experimental values. It was found that the varying the Smagorinsky constant affect the numerical simulation results. The predicted results with Pr = 0.1 have large deviation on results with both fine and coarse grids as compared with higher turbulent Prandtl number. Whereas, the effect of Sc and Cs on both fine and coarse grids has no significant effect on the predicted results. Since the buoyant plume in present experiment is not very large therefore the predicted results shows less dependency on the grid resolution. The effect of Smagorinsky sub-grid constant has less influence due to weak buoyant plumes.