The route choice of pedestrians during evacuation under conditions of both good and zero visibility is investigated using a group of experiments conducted in a classroom, and a microscopic pedestrian model with discrete space representation. Observation of the video recordings made during the experiments reveals several typical forms of behavior related to preference for destination, effect of capacity, interaction between pedestrians, following behavior and evacuation efficiency. Based on these forms of behavior, a microscopic pedestrian model with discrete space representation is developed. In the model, two algorithms are proposed to describe the movement of pedestrians to a destination under conditions of both good and zero visibility, respectively. Through numerical simulations, the ability of the model to reproduce the behavior observed in the experiments is verified. The study is helpful for devising evacuation schemes and in the design of internal layouts and exit arrangements in buildings that are similar to the classroom.

Route choice in pedestrian evacuation under conditions of good and zero visibility: Experimental and simulation results

We present a mobile lattice gas model for simulating the pedestrian evacuation process in a public building, through combining the advantages of the lattice gas model and the social force model. In our model, the interaction force between every two pedestrians and that between a pedestrian and the building wall are determined by the distance and the pedestrian’s moving step size. Simulation results show that the proposed model can capture the basic characteristics of pedestrian evacuation and generate the average evacuation times reasonably when an emergency incident occurs in the building.

A mobile lattice gas model for simulating pedestrian evacuation

The evacuation process of students from a classroom is investigated by both experiment and modeling. We investigate the video record of the pedestrian movement in the classroom, and find some typical characteristics of the evacuation, including variable velocity, dislocable queuing, monopolizing exit and so on. Based on the experimental observation, we improve the multi-grid model by considering the pre-movement time of each pedestrian, adopting variable velocity and using a new update procedure. With the improved multi-grid model, we simulate the evacuation process and compare the simulation results with the experimental results, and find that they agree with each other closely. In order to analyze the uncertainty of evacuation, we investigate the influences of the pre-movement time and its distribution on the evacuation. It is found that the evacuation times exhibit a (truncated) normal distribution and vary within a region of about 30% of the mean value. An interesting phenomenon is that the evacuation time of the egress experiment is close to the minimum value calculated with the model, due to the coordination among pedestrians during the experiment. The study may be useful in developing applicable egress models and understanding the basic egress behaviors.

Experiment and multi-grid modeling of evacuation from a classroom

The evacuation process in a teaching building with two neighboring exits is investigated by means of experiment and modeling. The basic parameters such as flow, density and velocity of pedestrians in the exit area are measured. The exit-selecting phenomenon in the experiment is analyzed, and it is found that pedestrians prefer selecting the closer exit even though the other exit is only a little far. In order to understand the phenomenon, we reproduce the experiment process with a modified biased random walk model, in which the preference of closer exit is achieved using the drift direction and the drift force. Our simulation results afford a calibrated value of the drift force, especially when it is 0.56, there is good agreement between the simulation results and the experimental results on the number of pedestrians selecting the closer exit, the average velocity through the exits, the cumulative distribution of the instantaneous velocity and the fundamental diagram of the flow through exits. According to the further simulation results, it is found that pedestrians tend to select the exit with shorter distance to them, especially when the people density is small or medium. But if the density is large enough, the flow rates of the two exits will become comparable because of the detour behaviors. It reflects the fact that a crowd of people may not be rational to optimize the usage of multi-exits, especially in an emergency.

Experiment and modeling of exit-selecting behaviors during a building evacuation

Modeling framework for optimal evacuation of large-scale crowded pedestrian facilities

Path Optimization Study for Vehicles Evacuation based on Dijkstra Algorithm

In this paper, evacuation dynamics in an office building is studied by experiment and simulation. A lattice gas (LG) model is developed. A parameter called ‘exit bias’ is introduced into the model to describe the occupants’ familiarity with different exits in a building. The evacuation experiment, which consists of seven scenarios under various conditions, is conducted to verify the model and calibrate the model’s input parameters such as pedestrian speed and exit bias. The effect of exit width on flow rate, and the effect of occupants’ familiarity with the building on their route selections, are studied. It is found that the accuracy of simulation depends a lot on the model’s pedestrian speed. The optimal pedestrian speed is decided by not only occupant characteristics, but also flow features determined by people distribution, building structure, environment pressure, etc. LG models with proper pedestrian speed are capable of simulating the dynamic process of orderly emergency evacuations.

Lattice gas simulation and experiment study of evacuation dynamics

 In this paper, small-grid analysis of discrete model is described, and simulation that some walkers leave a hall is carried out to check the effects of different desired walk velocities with the same walk time at a time step, and different numbers of small grid at a time step with the same desired walk velocity, on the evacuation time. The simulation results show that small-grid analysis have reproduced some typical phenomena of evacuation, including jam, block and faster-is-slower, etc. as good as the continuum model, i.e., the social force model, but with high simulation efficiency. In addition, the power-law distribution of evacuation flow duration and block duration with the different desired walk velocities is found. The block duration with different numbers of small grid at a time step also takes on power-law characteristics, only their intercepts in log–log coordinates are different.

Small-grid analysis of discrete model for evacuation from a hall

A behavior-based lattice-gas model for pedestrian dynamics is presented. This model adopts the behaviorism for mobile robot, and the walk task of pedestrian can be divided into three basic behaviors, i.e., “move”, “avoid”, and “swirl” basic behaviors. The walk direction is determined from the walk weight, which is the sum of the product of each vector of basic behavior multiplied by the weight in the corresponding direction. This model can simulate pedestrian movement with different walk velocities through the update at different time-step intervals. The periodic boundary for pedestrian counter flow with six simulation conditions in the channel is considered, and the dynamical characteristics are discussed. Simulation results show this presented behavior-based model can simulate some characteristics of pedestrian counter flow, e.g., lane formation and jammed configuration, etc. In addition, the different simulation conditions result in the different numbers of phases and their different critical total densities. In general, the mean flow 〈J〉 is always high if the corresponding mean velocity 〈V〉 is high, and their phases also turn at the same critical total density.

A behavior-based model for pedestrian counter flow

Pedestrian behavior models have successfully reproduced human movement in many situations. However, few studies focus on modeling human behavior in the context of terrorist attacks. Terrorist attacks commonly occur in crowded public areas and result in a large number of casualties. This paper proposes a three-stage model to reproduce a series of complex behaviors and decision-making processes at the onset of an attack, when pedestrians generally do not have clear targets and have to deal with fuzzy information from the attack. The first stage of the model builds a Bayesian belief network to represent the pedestrians’ initial judgment of the threat and their evacuation decisions. The second stage focuses on pedestrians’ global assessment of the situation through an analogy with diffusion processes. The third stage uses a cost function to reproduce the trade-offs of distance, safety, and emotional impact when considering a path to take. The model is validated using a video from the November 2015 Paris attack. The behavioral characteristics and trajectories of three pedestrians extracted from the video are reproduced by the simulation results based on the model. The research can be used to set rules when performing risk analysis and strategic defensive resource allocation of terrorist attacks using agent-based simulation methods.

Shifei Shen

Papers

Path Optimization Study for Vehicles Evacuation based on Dijkstra Algorithm

Lattice gas simulation and experiment study of evacuation dynamics

Small-grid analysis of discrete model for evacuation from a hall

A behavior-based model for pedestrian counter flow

A three-stage evacuation decision-making and behavior model for the onset of an attack