• Fire Science and Engineering
• /
• v.26 no.2
• /
• pp.59-68
• /
• 2012

Predicting occupants' behaviors from the start of the fire to egress and reducing the time required for such process are critical matters that can decide success and failure of safe egress. In this research, research literatures and theories and fire cases were compared and analyzed so as to prepare logical grounds that could predict the process of beginning egress. As a result of this research, there was a significant difference in the time elapsed until people start evacuating due to spatial positions and quarantine from the place from which the fire originated and their auditive and olfactory signs did not recognize the fire instantly and they showed a strong tendency to recognize the fire by visual sign, warning announcement for egress and notice by others. And the results also showed that only a very small minority of occupants evacuated as soon as they perceived the fire and that variation in the time elapsed until evacuation begun for occupants were wider as the size of building was bigger and that accommodations such as hotel had wider variation in the time elapsed regardless of the size of buildings.

• Fire Science and Engineering
• /
• v.28 no.6
• /
• pp.35-40
• /
• 2014

Bottleneck occurs as many people crowd into narrow doorway or corridor. Delaying egress time is occurred by bottleneck effect, and it is very important phenomenon on the egress analysis for building fire. An analysis of egress time should includes flow rate for considering bottleneck. Flow rate is numbers of people who pass the narrow gate as door or start point of corridor per unit length and unit time. The flow rate resulted from egress modeling should be approached to the result of experiments. In this study, flow rates from modeling by 'Pathfinder' and experiments was compared. The difference between the result from egress modeling and the one from experiments was verified. The average value of experiments is $4.25N/m{\cdot}s$, and the maximum average value of modeling is $1.55N/m{\cdot}s$.