Search

In 2017 in Tokyo, more than thousand occupants of a seven-story office building evacuated due to fire located at an adjacent site. Because the fire did not occur on their own site, the evacuation was spontaneous and, therefore, not performed according to a normal evacuation manual. This paper investigates the human behavior in this evacuation case, especially focusing on the decision-making process. To do this, the authors conducted face-to-face interviews with four people from the organization who made the decision to let employees evacuate and forty-one employees who evacuated. The major findings of the study are: (1) visual observation of fire through windows influences the occupants’ evacuation strategy; (2) occupants tended to postpone leaving the workplace until they received an explicit evacuation cue from others; and (3) multiple occupants chose stairs by taking into consideration that the fire was not in their building. Since the fire compartments had not yet been built and there was no immediate danger to life, these results can also be applied to the evacuation of non-fire floors in a building fire.

The effect of cross-wind on the temperature elevation in the downwind region of rectangular diffusion burners was investigated in wind tunnel experiments. The experimental parameters were the geometry of the diffusion burner (the side length parallel and orthogonal to the cross-wind L 1 = 0.2 – 0.8 m and L 2 = 0.4 – 2.0 m , respectively, and the aspect ratio r = 1 – 10 ), the heat release rate ( Q ˙ = 38.3 – 383.3 kW ), and the inflow velocity of the cross-wind ( U ∞ = 0.59 – 1.49 m / s ). Following the concept of the previously proposed point fire source model, the fractional contribution of the cross-wind velocity on the velocity along the trajectory N was introduced to close the system of governing equations, and a relevant line fire source model was derived. The line fire source model was correlated by the measurement results of the present experiment for the diffusion burner with the largest aspect ratio ( r = 10 ). Although the predicted temperature elevation along the trajectory by the present model was similar to that of the existing models, there was a substantial discrepancy in the predicted trajectory. For the unified analysis of temperature elevation and trajectories of the fire sources with various aspect ratios, the point fire source model was firstly tested to correlate the measurement results. As expected, the agreement was less successful for the results of the fire sources with large aspect ratios, especially in the region close to the fire source. Thus, the point fire source model was modified to consider the physical effects of the aspect ratio of the fire source and the attachment of the fire plume to the ground by invoking the correlations for the line fire source. With the modified model, the agreement between the predicted and measured results significantly improved. The proposed model is useful for assessing the hazards of wind-blown fire plumes in large outdoor fires on human activities such as firefighting or the evacuation of residents.

A considerable number of studies on emerged flames have been done in the past. However, since these studies were implemented using a single opening, the same conditions applied to multiple openings in close proximity to each other are yet to be determined.

The authors conducted an ISO 13785-2 modified large-scale facade fire tests in Tokyo University of Science and tentatively proposed the draft for the calculation method for vertical upstairs fire spreading based on those fire test results, prior to this study. And in this research, several modifications were carefully added to past calculation methods, especially regarding the calculation of temperature in the fireroom, detailed method of unifying multiple openings on the same floor for calculation, and the criteria for the occurrence of upstairs fire spreading. Detailed calculation methods are proposed for two different situations depending on the existence of an eave above the opening at the fireroom, respectively. Also, two different criteria for verification of vertical upstairs fire spreading were proposed, one of which is based on the maximum heat flux to the upstairs room (Method 1), while the other is based on the integrated value of heat flux over time of fire duration (Method 2). Using this newly modified calculation method, the effectiveness of 90 cm spandrel and 50 cm eave is demonstrated, respectively, in terms of practically preventing the occurrence of vertical upstairs fire spreading. Improved verification calculation methods (both Methods 1 and 2) are applied to four exiting office buildings in Japan and seven rooms of fire origin considered. Verification methods using calculation proposed in this study are proved to be able to practically evaluate the occurrence of vertical upstairs fire spreading, especially regarding noncombustible facades with openings in there.

Flame spread along eaves has been one of the major causes of fire spread for Japanese historic urban areas. For the fire safety assessment of such an environment, this paper investigates basic characteristics of heat transfer under eaves due to flame impingement. A series of model experiment was conducted measuring temperature rise and incident heat flux under nonflammable eaves with varying parameters on burner geometry, burner installation position, and fuel supply rate. Flow behavior under an eave was analyzed following the approach of the classical plume theory. The analysis result showed that the temperature rise is constant in the flaming regime and decreases proportionally to the distance to the power of −1 in the nonflaming regime. Reasonable correlation was obtained with the experimental results; the correlation was better when using the distance along trajectory from the burner as the reference length scale, rather than the distance from the impingement point.

Flame height, heat flux to the wall surface, centerline temperature rise, and mass flow rate from a line burner along a vertical wall were measured to propose an experimental correlation for mass flow rate previously not established. The heat release rates and burner heights were changed for three burners of lengths 0.2 m, 0.3 m, and 0.5 m. The intermittent flame heights and heat flux to the wall surface were consistent with experimental correlations of an earlier study by Hasemi. The centerline temperature rise in plume region was inversely proportional to the distance from the fire source, consistent with the theoretical analysis. The horizontal temperature distribution in the plume region was consistent with a Gaussian distribution. The mass flow rate was proportional to the distance from the fire source. The entrainment coefficient appears to be consistent with earlier studies for line plumes in an open space.

To evaluate the importance of fire protection systems objectively in evacuation safety planning of building fires, authors have proposed a methodology for selecting both design fire scenarios and design fire simultaneously for performance-based fire safety design according to acceptable evacuation risk in fire, which we call risk-based evacuation safety design method or R-B ESDM in abbreviated term. An important factor when using the R-B ESDM is to set the reliability of sprinkler systems. Although several studies exist on the reliability of sprinkler systems, there is no study that looks into the relations of design fire and design fire scenarios. In order to introduce the reliability of sprinkler systems to R-B ESDM, the data of the national fire reports have been statistically analyzed by classifying the fires in the sprinklered spaces into three patterns: “fires that sprinklers extinguished,” “fires that sprinklers suppressed,” and “fires that sprinklers failed to control.” The probabilities of the three patterns have been calculated by using the number of fires according to the burned area. As a result, if the burned area for effective suppression is defined as less than 4 m2, the reliability of effective sprinkler activation turns out to be approximately 83 %.