Fire Emergency Crowd Evacuation Simulation Using Multi-Grid Model

In combination with Mechanics, psychology, sociology,etc, crowd simulationbased on computer algorithms isused to show the motion process of the crowd underthe computer-generated virtual environment,studying the motion characteristics andmovement law of the crowd in various environment.Crowd simulation which providesguidelines to building safety and emergency plans is a significant way to preventand reduce injures and death during emergencies such as fires.Cellular Automata model and Social Force model, the most typical models incrowd evacuation, have their own disadvantages:Cellular Automata model can notproperly demonstrate interactive forces among individuals,thus causing errors.Inaddition,this model can not accurately estimate influence of emergencies,like fires,onthe crowd.Meanwhile,Social Force model is so complicated and lacksefficiency.Under this circumstance,this paper proposes a new evacuation model basedon law of force in fire environment which is inspired by Multi-grid model. Its maincharacteristics include:(1) Adopting fine grid to divide the space, multi-grid is good atdescribing subtle movementsof the crowd.Boundary conditions can also be accuratelyquantified.(2) This model includes interactive forces when calculating locomotivedirections.Similar to Social Force model,this model can sum up interactive forcesthrough identifying the amount and locations of grids occupied.(3) This modelnotonly adopts the drift and the interaction force, but also brings in firerepulsion,demonstrating the impact of fire on pedestrians and even more vividlysimulating the details of crowd evacuation.(4) This model uses a global gridalgorithm to calculate collision.Its computational efficiency is as better as that ofCellular Automata model,but much higher than that of Social Force model.Based on the model proposed above, this paper constructs an original simulation system of crowd evacuation which can be applied to fire emergency.A series ofparameters can be set and simulation can be dynamically observed and compared. Thelast part of this paper discusses the impact of self-driving force, friction and firerepulsion parameters on the simulation, and uses this model to re-stimulate sometypical self-organization phenomena, such as the arched distributionof evacuation atthe bottleneck, congestion, intermittent flows of people and situations of pedestrianskeeping off the danger zones.Finally,this paper analyzes the simulation of crowdevacuation in terms of fire repulsions,fire spreading levels and fire locations.