For the building evacuation in the case of fire or earthquake, people’s visual field can be usually obstructed by the resulting smoke, or the failure of the electrical power supply. Due to the reduction of the visual field, evacuees may become disoriented and fail to escape. Imagine that, if the disoriented evacuees could be guided by some experienced evacuation leaders that were familiar with the internal layout of building and could recognize the escape route to exit during evacuation, the evacuation time would maybe reduced significantly.
This study aims to address major issues for the leadership in evacuation: what are the effects of leadership in evacuation? How to realize the effective leadership for crowd evacuation? To address these issues, a series of simulations under different evacuation conditions (e.g., different visibility conditions, different sizes of crowd, and different number of leaders) based on the developed social force model in the typical rooms were conducted. The results showed that:
Firstly, when the visibility condition of room is very limited, evacuation time can always decrease with the increase in the number of leaders regardless of the size of crowd. And the smaller the size of crowd, the larger the number of leaders is required to assist evacuation for whole crowd effectively.
Secondly, when the visibility condition of room is large enough, the increase in the number of leaders cannot always lead to the decrease in the evacuation time; on the contrary, a few of leaders in a crowd may increase the evacuation time and slow the evacuation efficiency.
These two findings indicated that leadership can accelerate evacuation and play a positive role when the visibility condition of room is very limited. They also revealed the existence of a possible negative effect of leadership and that the leadership effect may transit from positive to negative when the visibility condition of room is improved.
In addition, we also discovered that the effects of the recognizable leaders are superior to the unrecognizable leaders when the visibility condition of room is relatively limited; on the contrary, the effects of the unrecognizable leaders may be superior to the recognizable leaders when the visibility condition of room is large enough.
To verify these interesting findings from simulation, a series of the laboratory experiments under different evacuation conditions were conducted with a group of 15 participants in a built evacuation scenario. The experiment results successfully verified the transition of the leadership effect from positive to negative, and the superiority comparison results between the recognizable and unrecognizable leaders. Besides, from the experiments, three different navigation patterns of evacuees for going out of the room were revealed, namely, either following the wall, following the crowd stream or following the detected leader. It was also found that the selection on the navigation pattern is determined by evacuation conditions, such as, the visibility condition of room, and the number of leaders.
To sum up, this thesis on the one hand developed a modified social force model. It could be used to investigate the effects of the leadership in the evacuation under limited visibility and other evacuation dynamics in many other contexts, on the other hand proposed an ingenious but easy-conducted experimental methodology. It could be used to investigate evacuation dynamics under limited visibility. This thesis also provided the investigation results about the effects of the leadership in evacuation from the perspective of simulation and experiment. The obtained findings, results and conclusions from this thesis are important basis about how to carry out leadership in evacuation to assist evacuation effectively.