Crowd Evacuation: The Social Force Model and Faster-Is-Slower

People as particles with a will of their own

The Social Force Model, introduced by Dirk Helbing and Péter Molnár in 1995, treats each pedestrian as a particle with mass and velocity that is nonetheless driven by psychology, not just physics. Every person i feels a sum of forces that Newton's second law then turns into acceleration:

m_i · dv_i/dt = F_drive + F_pedestrians + F_walls + F_random

F_drive pulls each person toward their desired exit at their preferred walking speed. F_pedestrians and F_walls are repulsive — they grow sharply as people or obstacles get close, keeping personal space and preventing overlap, typically modelled as an exponentially decaying force plus a short-range hard-body term once bodies actually touch. A small random term captures the fact that people are not perfectly rational agents.

Bottlenecks and the arch that forms at doorways

Push enough agents toward a single narrow exit and a striking, repeatable pattern appears: people do not funnel through smoothly, they self-organize into an arch or wedge shape a few body-widths back from the opening, with pressure building at its apex. This is not scripted — it falls directly out of the repulsion terms, and it matches video of real crowd evacuations closely enough that architects use social-force models to certify exit widths before a building is built.

Faster-is-slower

The model reproduces one of the most counter-intuitive and well-documented findings in pedestrian dynamics: past a critical panic level, increasing everyone's desired speed decreases the exit flow rate. As agents push harder toward the door, the repulsive forces near the bottleneck spike, bodies jam together, occasional clogs and "stop-and-go" arching form, and the net throughput of people per second through the doorway drops below what a calmer crowd would achieve. Simulations and controlled experiments with real crowds agree on this effect, which is why evacuation guidance stresses orderly movement over pure speed.

Why removing a small obstacle can speed evacuation

A second counter-intuitive result: placing a small column or barrier slightly off-centre in front of a doorway can raise throughput, because it splits the pressure of the arch into two smaller, more stable clusters instead of one large chaotic pile-up. The effect is fragile — the obstacle has to be sized and placed correctly, or it just becomes another thing to bottleneck around — but it is a real, model-and-experiment-confirmed design trick used in stadium and stair exit planning.

What the parameters mean

Turning up the desired speed raises everyone's target walking speed, which is where faster-is-slower kicks in above a threshold. The relaxation time controls how quickly a person's actual velocity catches up to their desired one — smaller values mean sharper, more reactive movement. The repulsion strength and range set how much personal space people insist on and how strongly walls push back; turn repulsion down and you get unrealistic overlapping; turn it up too far and small gaps freeze into gridlock. Exit width is the single most sensitive geometric parameter: throughput scales roughly linearly with width once it clears about one shoulder-width per lane, but narrower openings suffer disproportionately from arching.

Frequently asked questions

Try it live

Everything above runs in your browser — open Evacuation and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.