This work presents a model to predict the instantaneous rate of convective heat transfer from fire plume gases to the overhead ceiling surface in a room of fire origin. The room is assumed to be a rectangular parallelepiped and, at times of interest, ceiling temperatures are simulated as being uniform. Also presented is an estimate of the convective heat transfer, due to ceiling-jet-driven wall flows, to both the upper and lower portions of the walls. The effect on the heat transfer of the location of the fire within the room is taken into account. Finally presented is a model of the velocity and temperature distributions in the ceiling jet. The model equations were used to develop an algorithm and associated modular computer subroutine to carry out the indicated heat transfer calculations. The subroutine is written in FORTRAN 77 and called CEILHT. The algorithm and subroutine are suitable for use in two-layer zone-type compartment fire model computer codes. The subroutine was tested for a variety of fire environments involving a 107 W fire in a 8m × 8m × 4m high enclosure. While the calculated results were plausible, it is important to point out that CEILHT simulations have not been exprerimentally validated.
Two series of experiments have been carried out on the effect of roof vents on the operation of sprinklers in a building approximately 50 m × 20 m × 10 m high. This paper describes experiments in which measurements were made of temperature and velocity distributions produced by a 2 m square hexane fire with a controlled theoretical heat output of 5 MW. Combinations of 0, 10 or 20 vents, each having an area of 1.64 m2, and 0, 1 or 5 sprinklers were investigated. The plume was generally deflected from the vertical and estimates were made of the position of the centre of the plume at ceiling level. The deflection was increased by the sprinkler spray but was not increased by venting. The results have been compared with the distributions calculated by a previously-described simple zone model of the effects of venting on sprinkler operation with a growing fire. The model makes a good approximation for the radial temperature distribution beyond the 'turning region' of the plume when there are no sprinklers. Within the turning region the assumption of a temperature plateau is valid when there is a deep layer of hot gases but not otherwise. With the vents used, local effects of vents on temperatures 100 mm beneath the ceiling were small compared with the 'global' effects calculated by the model. The model generally underestimates the maximum velocity of the ceiling jet but the mean velocity is predicted more closely. The model requires better information on the cooling of the hot gases by the sprinkler spray. There is some evidence that cooling by the water droplets may not in itself be sufficient to explain the observed temperatures and that there may be some additional cooling. This may be by increased entrainment of hot gases into the ceiling jet caused by the sprinkler spray. The work provides evidence that calculations using the model overestimate the effect of venting on the operation of the first sprinkler and may underestimate the reduction by venting of the number of sprinklers operating beyond the turning region of the fire plume.
Experiments on the operation of sprinklers with and without permanently open roof vents have been carried out in a building approximately 50 m × 20 m × 10 m high. They were intended to validate a simple mathematical model to calculate the time of operation of sprinklers given the rates of growth of the fire before and after operation of sprinklers as an input. The fire was produced by burning hexane floating on water and grew exponentially with a time constant of 60 s until the first sprinkler operated. Thereafter the heat output remained constant; in some experiments it was reduced by 20% some 30 s after the first sprinkler operated. The first sprinkler operated when the heat output was of the order of 10 MW; the delay due to venting was about half that calculated using a simple model described in a previous report. If vents were closed or the water supply was poor then all or most of the available sprinklers operated but if vents were open and there was a good water supply then a limited number of sprinklers operated. The simple model did not predict that venting could reduce the possibility of a large number of sprinklers operating after the fire was controlled. The experiments confirmed the results of calculations using the model, that the prior opening of vents would have little effect on the operation of the first sprinkler. They also indicated that prior venting would not increase the number of sprinklers opening after control of the fire and might reduce it; when the fire growth process is completely independent of the number of sprinklers operating.
Experience in using state-of-the-art zone type fire models for analysis of fire tests and the design of smoke control systems for large buildings is discussed. In one case, correcting for the sideways blowing of the fire plume by the jet of air entering the fire room changed the qualitative fire behavior. The modeling of horizontal vents between floors of a nearly closed building presents a challenge for current models. It can only be satisfactorily met for special cases; others may require assumptions than can not be strictly justified. Modeling large area low ceiling rooms can be improved by the use of "pseudo rooms" though this technique is largely unsupported by experimental data. High ceiling, large area rooms - atria - are not well treated by present zone models.
Full scale burning tests were performed to study wood, polymethylmethacrylate (PMMA) and methanol fires in a compartment under forced ventilation conditions. In each fire test, the mass loss rate, the gas temperature, the flame height, carbon monoxide, carbon dioxide and oxygen concentrations were measured. Results illustrated that the burning behaviour of wood will be affected by the ventilation rate. But for PMMA and methanol fires, the burning behaviour and the fire plume properties are not so sensitive to the change in ventilation rates. The hot gas temperatures for wood and PMMA fires within the quasi-steady burning period are expressed in terms of the extraction rates and the average total heat-release rate with empirical constants determined by this experiment. Also the average hot gas temperatures are compared with the values calculated by the simple model due to Deal and Beyler (1990).