Modeling Of Pedestrian Evacuation Dynamics Based On The Heterogeneous Lattice Gas Model

Evacuation under emergency is a complex process. Due to the limits of undertaking such dangerous experiments and the absence of data from real evacuation, computer simulation is the main approach for studying emergency evacuation at present. Two kinds of models, i.e. continuous models (the social force model) or discrete models (the cellular automata model) are widely used for simulating pedestrian evacuation. In a social force model, for N pedestrians, the requirement of solving a couple of differential equations for each pedestrian implying that the calculation load is the order O(N2), which is a heavy computation burden for large N. It means that the computational cost is very expensive in general. By contrast, a cellular automata model can greatly improve the computational efficiency in simulating large-scale pedestrian evacuation since the calculation amount is the order O(N). The interaction between pedestrians which is a fundamental element plays a key role for simulating pedestrian evacuation.But in a cellular automata model, the interaction between pedestrians is neglected.Aiming at the problems mentioned above, the main contents and achievements in this paper are listed as follows:(1) Based on the cellular automata method (CA model) and the mobile lattice gas model (MLG model), we have developed a heterogeneous lattice gas model for simulating pedestrian evacuation processes under emergency. A local population density concept is introduced first. The update rule in the new model depends on the local population density and the exit crowded degree. The drift D which is one of the key parameters influencing the evacuation process is allowed to change according to the local population density of the pedestrians. Interactions including attraction, repulsion and friction between every two pedestrians and those between a pedestrian and the building wall are described by a nonlinear function of the corresponding distance, and the repulsion forces increase sharply as the distances get small. A critical force of injury is introduced into the model, and its effects on the evacuation process are investigated. The model proposed has the heterogeneous features as compared to the MLG model or the basic CA model. Numerical examples show that the model proposed can capture the basic features of pedestrian evacuation, such as clogging and arching phenomena.(2) An extended heterogeneous lattice gas (E-HLG) model is developed by introducing an altitude factor into the heterogeneous lattice gas model (HLG model). The altitude factor is used to describe the position height of lattice sites. Evacuation features from a terrace classroom are investigated through both simulations using the model and experiments. To study evacuation processes under fire emergency, an agent-based fire and pedestrian interaction model (FPI model) is proposed. It is supposed that the possible moving directions of a pedestrian depend on the environmental temperature field which is simulated by the software FDS. The walking speed reduction due to the visibility worsening in the FPI model is described by a multi-grid method. It is found that simulation results based on the extended HLG model are in good agreement with the experiments. The altitude factor plays a guidance role to the evacuation, and the fire notably delays the evacuation due to both the harmfulness of the high temperature field and the change of evacuation routes which results in frequent local jamming and clogging.(3) An extended heterogeneous lattice gas model is developed by introducing an altitude factor into the heterogeneous lattice gas model for simulating pedestrian evacuation form subway station. Taking into account the mobile characteristics of pedestrians, the multi-grid method is introduced into the model. The walk speed is determined by the real-time moving step sizes of themselves. The influences of the initial distribution of pedestrians on evacuation process are investigated. The role of emergency exits to evacuation is analyzed. Evacuation processes of pedestrians with exit preference and under fire emergency are simulated by using the model mentioned above. Numerical examples show that moderate use of emergency exits can improve the evacuation efficiency and the altitude factor contributes to evacuation process.