STUDIES ON COMPARTMENT FIRES

Abstract

Fire research today is aimed to build up a predictive methodology to be able to forecast the circumstances and conditions under which a fire may grow, and also impact of fire safety measures on fire. In the earlier period, efforts were dedicated to determine the basic principles of fire growth, to measure the variables involved, and to develop coordinated approach to predict the course of fire. Several useful theories were evolved during this period. The objectives were concentrated more to know the basic mechanism of fire growth and spread, and flashover. Two important criteria for determining flashover, e.g. temperature of hot layer and the incident heat flux on other combustibles in the room, have been proposed and accepted. With the advent of fast computing machines, research works were later devoted to the development of appropriate mathematical models of fire in enclosures. In the present study, an effort has been made to generate data on fire growth in the laboratory and to evolve a simple yet effective pre-flashover zone model to predict the effects of a fire in an enclosure. The work has been carried out under three main headings as given below. Experimental studies: Experiments have been conducted in a small scale compartment / enclosure to generate data on temperatures versus fire strength. Development of zone model: A simple yet effective pre-flashover zone model has been developed to simulate fires in compartment / enclosure. Model validation: The suitability of the developed model has been determined by comparing its predictions with the experimental data obtained as well as with those available in the literature. Experimental Study The experimental setup comprises a compartment constructed with fired clay bricks in cement-sand mortar in 1:6 ratio. The floor area of the compartment is 1.76 m x 1.26 m and height is 1.06 m. Thickness of walls is 0.125 m. A rectangular opening, 1.22 m wide and 0.70 m in height, is provided in the front wall of the compartment. The interior wall and the ceiling surfaces are plastered with 0.013 m thick cement-sand mortar of 1:4 ratio. The compartment has been instrumented to monitor fuel burning rate, temperatures of hot layer and outgoing gases.The fuel chosen in this study is methanol and is burned in a square metai tray. Three different size trays, of area 0.01, 0.0225, and 0.04 m2, have been used for carrying out experiments. Measurements of temperatures at fourteen locations have been taken. Flame heights have also been measured with the heip of a cathetometer. Measurements of flame height have been taken within and outside the compartment. The study has revealed that the temperature of the hot gas layer formed beneath the ceiling, is not uniform throughout. Three distinct zones of temperatures have been observed. Temperatures just above the fire source and in its close proximity are found to be the highest. Room corners far away from the fire source are the coolest one. Temperature of the outgoing gases is found closer to that of the intermediate zone. Temperatures have also been recorded at three locations at the floor level. It is found that there is no significant rise in temperature at the floor level. Measured flame heights have been compared with its predictions by several correlations available in the literature. It is found that the compartment boundaries are not making any significant effect on the flame size. Relationships due to McCaffrey (1979), and Hasemi and Tokunaga (1984) provide the excellent comparison while other relationships provide much lower values of flame heights. Development of Mathematical Model A two layer pre-flashover model to predict temperature, interface height and neutral plane height in compartment has been developed. The model equations have been derived by the application of principles of conservation of mass and energy in the upper as well as lower layers. The model equations have been programmed in FORTRAN IV and solved with the help of fourth order Runge-Kutta method. Based upon the developed model, a user friendly software has been developed; this is named as CALTREE (CALculate Temperatures for Risk Estimation in Enclosures). Model Validation Predictions of temperatures by CALTREE were compared with the experimental data of Steckler et al. (1982), and those obtained in the present study. It is observed that the CALTREE has predicted comparable values of temperatures and interface locations for all experiments of Steckler et al. (1982), simulated in this study. On the other hand, the measured average values of upper layer temperatures obtained in the present study are in close agreement with the model predictions. Further, predictions of CALTREE have also been compared with the experimental data of Mowrer and Williamson (1987), and Dembsey et al. (1995). It is observed that the CALTREE has predicted temperatures very close to experimental values of Mowrer and Williamson (1987). CALTREE predicted higher temperatures for all the experiments of Dembsey et al. (1995). The reason for the deviation may be due to the point source plume model used in the development of CALTREE model. In order to improve its predictions further, Heskestad's finite size fire source plume model (1984) has been incorporated in CALTREE. The heat loss through boundaries has been calculated by incorporating lumped heat transfer coefficient concept in place of the fixed value of Xc. This has resulted in considerable improvement in simulations. The comparison of CALTREE model predictions with the experimental values reported by Steckler et ai. (1982), Mowrer and Williamson (1987), and Dembsey et al. (1995) has revealed that CALTREE provides excellent simulations of free burn fires in enclosures. CALTREE provides poor simulations for fires in corner and for the fires along the wall. This is due to the reason that the available plume models do not address the problem of restricted entrapment due to walls and corners. Mowrer and Williamson (1987) resolved this problem by applying the concept of multiplication factors for corners as well as the side wall fires. Mowrer and Williamson (1987) suggested that the excess temperatures predicted by McCaffrey et ai. (1982) formula, if multiplied by 1.7, will provide satisfactory simulations for corner fires. Similarly, multiplication factor of 1.3 is found suitable for side wall fires. CALTREE predictions, when multiplied by a factor of 1.65 for corner fires and 1.36 for side wall fires, became very close to the experimental values due to Steckler et al. (1982). Therefore, it may be concluded that these factors range between 1.65 and 1.75 for corners fires, and between 1.3 and 1.4 for side wall fires. It is our view that the CALTREE may be used satisfactorily for simulating fires during pre-flashover period. It has provision to accomodate finite size fire sources in addition to point sources. Further, the experimental data, generated during the course of present research work, shall be useful to validate other zone or field models, and also to estimate their parameters.