Experimental And Simulation Research On Pedestrian Dynamics In Bottlenecks

Human stampede is one of the major accident disasters in public gathering places. Bottlenecks become a high-incidence area of human stampede because of its intense contraction of spatial structure and tremendous contrast of capacity. In this work, pedestrian dynamics in bottlenecks was investigated by both experiments and simulations. The effects of various influencing factors on pedestrian dynamics were comprehensively analysed. The mechanisms of collective phenomena during evacuation process were explained. Reference bases for the optimal design of bottlenecks were proposed. The main points of this work are as follows.(1) A detection method of steady state in pedestrian dynamics was proposed. Based on the original CUSUM (Cumulative Sum Control Chart)algorithm in statistical physics, reasonable modifications were made according to characteristics of pedestrian dynamics. The detection threshold was calibrated by an autoregressive process and the stability of the modified CUSUM algorithm was comprehensively verified. The detection method lays a foundation for the following and future research.(2) Experimental research on pedestrian dynamics in fundamental bottlenecks was performed. Four groups of pedestrian experiments in fundamental bottlenecks were introduced. The collection, extraction and analysis of experimental data were illustrated. Pedestrian characteristics in experiments were comprehensively investigated. Combing the four groups of experiments and summarizing all other available pedestrian experiments in fundamental bottlenecks, their data were compared to draw universal conclusions. Main conclusions include the following. Pedestrian flows in wide and narrow bottlenecks follow the same variation rule. If the ratio of pedestrian number to bottleneck width is larger than 115 persons/m, the difference of pedestrian flows in all states and steady state could be ignored.Pedestrian flow shows a linear dependence on bottleneck width in both all states and steady state, and the slope of the dependence shows different values under the combined effects of various influencing factors. The ratio of corridor width to bottleneck width makes the slope of the dependence firstly increase then decrease and finally keep constant.(3) Research on validation of pedestrian model based on pedestrian experiments was performed. The pedestrian model FDS+Evac was used to simulate pedestrian dynamics in fundamental bottlenecks. A hybrid validation exercise of simulation results based on experimental data was conducted with a comprehensive combination of multiple pedestrian characteristics. The hybrid validation substantially enhances the reliability of simulation results.(4) Research on human route choice in complex bottlenecks was performed. Three groups of pedestrian route choice experiments in complex bottlenecks were introduced. The influences of exit width, the distance to exits and the crowd density around exits on initial route choice behaviour and path re-planning behaviour were investigated. The intrinsic mechanism of human route choice was illustrated. Main conclusions include the following.Initial route choice behaviour and path re-planning behaviour are responsible for balancing exit usage with leading and assisting roles, respectively.Anticipation of initial route choice behaviour is conducted based on exit width, the distance to exits and the crowd density around exits. Path re-planning behaviour is universal but of small probability, and the distribution for occurrence ratio of path re-planning behaviour is similar to a normal distribution.(5) Research on layout optimization of complex bottlenecks was performed. A route choice model based on cellular automata model was established and used to simulate pedestrian dynamics in three complex bottlenecks with different arrangements of exits. The distance to exits and the crowd density around exits were considered in route choice. The layout optimization for arrangement of exits in complex bottlenecks was recommended.The innovations of this work are as follows. (1) A detection method of steady state was specially proposed for pedestrian dynamics for the first time,which makes related research normative. (2) A pedestrian experiment in wide bottlenecks was performed, which enlarges the investigation of bottleneck width to 5.0 m. (3) A comprehensive hybrid validation method with a combination of multiple pedestrian characteristics was proposed, which substantially enhances the reliability of validation results.The research on pedestrian dynamics in bottlenecks is the key point for preventing human stampede. This work provides scientific and reasonable reference bases for the optimal design of bottleneck scenarios, helps to reduce the incidence of human stampede, and improves safety management work in public gathering places.