Modified Social Force Model Based on Predictive Collision Avoidance Considering Degree of Competitiveness

Modeling building evacuation during fire emergencies is an important issue. The social force model is a well-regarded evacuation modeling technique, and it has been integrated into the Fire Dynamics Simulator with Evacuation (FDS + Evac) of NIST to simulate building fire evacuation. However, these models still have limitations to be improved. First, occupants’ movement can be unrealistically prevented. For example, the corner of doors exerts unrealistic repulsive forces on occupants, so the simulated flow at narrow doors is much smaller than experimental data. Second, the degree of occupants’ competitiveness is not considered. Finally, current models are rarely validated by data form real-life emergencies, in which occupants may behave differently from normal situations. In this paper, new social forces are used to replace old ones to modify occupants’ collision avoidance: both time headway and time-to-collision are viewed as indicators of potential collisions, and social forces are active if time headway or time-to-collision reaches thresholds. A parameter is used to represent how the degree of occupants’ competitiveness affects their collision avoidance. The modified model is validated by both lab experiments and real emergency evacuation. First, the relation between simulated flow and door width in non-competitive situation is used for validation. In the simulation, 94 occupants, initially distributed in a 9 m - by - 4 m area, evacuate from a door. The simulated flow rates through doors of width ranging from 0.6 m to 1.2 m are consistent with the experimental data. Second, effects of competitiveness are studied. Simulation results show that whether competitiveness speeds up or slow down the evacuation through a door is affected by the initial number of occupants, door width, and other occupants outside the door. Finally, simulation results in competitive situation are consistent with data from a real-life emergency evacuation. The data used is extracted from a video recording occupants evacuating from an airport through a security gate in an earthquake. Simulation results are consistent with the real-life data in both the total evacuation time and the time when congestion occurred.