A Study on Pedestrian and Evacuation Dynamics Involving Human Behavior

考慮人員行為特征的行人與疏散動力學研究

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date29 Dec 2017

Abstract

    Pedestrian evacuation is an important branch of safety research. It is also one of the most effective ways to ensure human safety under emergency situations. Research on pedestrian and evacuation dynamics is helpful in efficient, safe and comfortable operating environment design (e.g., subway stations, stadiums and airports) and large-scale crowd management. With the increasing number of unexpected events in recent years, pedestrian and evacuation dynamics has drawn much interest and attention. A number of models and experiments have been conducted to analyze crowd behavior and evacuation dynamics. However, the combination of pedestrian movement and psychological influence especially in cellular based models has not been well investigated, along with the effect of the surrounding environment on pedestrians’ decision making during evacuation processes.

     In this thesis, we first study the combination of pedestrian movement and psychological influence through modeling. As emotion plays an important role in individuals’ decision making in some emergency situations, and the contagion of emotion may induce either normal or abnormal consolidated crowd behavior, we seek to simulate the dynamics of emotional contagion among crowds by modifying the epidemiological SIR (susceptible–infected–recovered) model to a cellular automaton approach. According to SIRS contagion in epidemiological theory, a new cellular automaton model (namely CA-SIRS model) is established to capture the dynamic process, i.e., “susceptible–infected–recovered–susceptible”. This process is integrated with individual movement. Simulation results of this model demonstrate that multiple waves and dynamical stability around a mean value will emerge during emotion spreading. It was found that the proportion of initial infected individuals had little influence on the final stable proportion of the infected population in a given system, and that infection frequency increased with an increase in the average crowd density. Our results further suggest that individual movement promotes emotion spread and increases the stable proportion of the infected population. A decrease in the duration of an infection and probability of reinfection can evidently reduce the number of infected individuals. According to these findings, we employ a multi-grid model to understand pedestrian dynamics in counter flow, coupled with the effect of emotion propagation in a crowd stampede induced by panic. The time evolution of average speed under conditions with and without emotion propagation is obtained. It is observed that though emotion propagation increases the desired speed, it easily results in congestion due to an increase in competition in counter flow. According to the speed–density relationship, the crowd density and percentage of pedestrians in counter flow are used to discuss the flowability of pedestrians in a long channel. Thus, three regimes, i.e., lane formation, the transition stage and clogging, are achieved. This is important for the organization of safe mass events concerning crowd density. The force distribution and cumulative distribution of the largest force that pedestrians experience are further presented to understand a tragic incident related to crushing such as the Cambodian stampede in 2010. These results provide an estimation of the number of injuries from the perspective of force.

     Then we seek to explore pedestrian evacuation processes, given the influence of the surrounding environment in buildings, such as emergency signage and surrounding pedestrians or obstacles, with experimental and modeling methods. Experiments were undertaken in a building, given individual and collective movement in two scenarios. 119 individuals participated in 9 tests. Different instructions were given to them, leading to 4 movement conditions. The whole experimental process was recorded by DV cameras and head mounted mini video cameras. Exit choice, way finding behavior, the effect of emergency signage and decision time are discussed according to post-trial questionnaires and video recordings. Results demonstrate that most of participants are inclined to choose the nearest exit, regardless of their unfamiliarity with the building layout. Herding is frequently discovered during the crowd movement. In scenario 1, both the average signage detection probability and average probability of following the sign information under individual situations are higher than those under crowd situations. Few participants pay attention to low placed exit signs in different conditions. Furthermore, average decision time for familiar and unfamiliar participants at different decision points is compared. According to experimental findings, a discrete evacuation model defined on a cellular space is introduced according to the fuzzy theory which is suitable to depict imprecise and subjective information from the surrounding environment. Pedestrians’ perception-based information and different characteristics are used as fuzzy input. Then fuzzy inference systems with rule bases, which resemble human reasoning, are built to achieve fuzzy output that governs pedestrians’ movement direction. This model is performed under two scenarios, i.e., in a single-exit room with and without obstacles. Simulation results reproduce some typical phenomena (e.g., arching, clogging and the “faster-is-slower effect”) discovered in real building evacuation situations, and are consistent with those in other models and experiments. This proposed method is helpful to enrich movement rules and approaches in classic cellular based evacuation models.

    According to the fuzzy inference technique, the transmission of information on danger that has an influence on crowd dynamics during the process of crowd dispersion is further investigated. Fuzzy functions and rules are established for the vague description of human states and information received. Fuzzy inference is conducted to obtain the output values of decision making, e.g., pedestrian movement speed and directions. Simulation is performed under a four-way pedestrian situation. Good crowd dispersion phenomena are gained. Simulation results under different conditions demonstrate that information transmission cannot always result in successful crowd dispersion under all conditions. This is dependent on whether decision strategies in response to information on danger are unified and effective, especially in dense crowds. Results also show that increasing the strength of the drift at low density or the proportion of pedestrians who select one of the furthest unoccupied Von Neumann neighbors from the dangerous source as the drift direction at high density is useful in crowd dispersion. Compared with previous work, this study improves an in-depth understanding of nonlinear crowd dynamics under the effect of information on danger.

    Finally, conclusions are presented. Contributions are also emphasized. Accordingly, recommendations for further research studies are stated.

    Research areas

  • Pedestrians, Evacuation, Emotional contagion, Human behavior, Surrounding environment, Fuzzy theory, Information transmission, Cellular based models, Experiments