To understand the fire characteristics of multiple fire sources in three dimensions, we conducted a series of experiments using two small pool fires. As a result, mass burning rates for both pool fires are larger than that of a single pool fire, especially the mass burning rate for the upper pan remarkably increases compared with that for the lower pan. In this paper, the effects of external radiation and buoyant flow on mass burning rate were analytically examined. The external radiative heat flux was calculated using a cylindrical flame model. Additional convective heat flux due to induced buoyant flow by other fire source was estimated by modified Grashof number. The calculation results comparatively consist with experimental results. The calculation model was applied to examine the scale effect of fire sources for the upper pan. Additional convective heat transfer rate is larger than that of external radiation in small scale fire sources less than 0.2 m of diameter, while the external radiation becomes to be dominant heat transfer mode in fire sources above 0.5 m of diameter.
Author tries to construct the frame in which we can discuss the optimality of fire-safety evaluation system. At first, introducing a simple model on the gain and loss produced by the judgment of fire-safety evaluation system, it is shown that the desired evaluation system should minimize the loss from the result of the judgment. Next, a requirement is deduced. The requirement expresses the optimal condition of system by three factors : 1) conditional probability of observation data under the each fire safety level, 2) appearance probability of the building of the each safe level and 3) the loss by each case. In addition, a proposition that expresses the requirement of the error smallest evaluation system is obtained. Finally, point addition type evaluation system used by actual evaluation system was examined. It is shown that although the point addition type evaluation system is very simple, it has the possibility to become the loss smallest evaluation system when each item is statistically independent.
Until recently, validity of simulations for a subway station fire was difficult to be evaluated as there was almost no available data for a real subway station fire. To obtain some real data, in October 2003 in Tokyo, smoke movement experiments were conducted in actual subway stations currently in service. Although this experiment was limited in time, intensity and contents, essential data was collected to tune and examine the validity of the numerical calculations. The experimental results suggested that the smoke movement is highly dependent on the air flow blowing down the stairway and the opening and closing conditions of the shutters. The reasonable agreement between the experimental data and the prediction shows that the CFD model can reproduce the smoke- air mixing which is caused by the air flow through the stairways connecting the platform and the concourse. The CFD model will be suitable to reproduce the specific smoke movements in subway stations. Smoke spread at the event of more plausible severer fires in more complicated conditions is discussed to grasp fire safety levels and to propose appropriate operation of shutters and smoke extraction systems.