Evacuating a building in an emergency is a critical operation, and many lives can be lost in the process.

One famous example is on 9/11: “between 8:46 AM and 9:03 AM, an estimated 1,400 people successfully evacuated the upper floors of the South Tower, while roughly 600 people did not.”

Being able to predict and optimize evacuation routes will hopefully help save lives in emergencies, or at least expedite the process of evacuation.

When Niles West holds a practice evacuation, students often walk at a slower pace than they would when they walk to class; this causes backups in the hallways and slows everyone down. If one person was alone in the building, they would leave a lot more quickly.

Although this speed is likely unachievable with a crowd, I hope to figure out how to help buildings evacuate with speed close to that of an individual in egress.

Concepts

• ​Using adjacency matrices and network diagrams for egress(exiting) analysis. Variables in the emergency/in general (where the fire originated, exit location/size).
   

• ​Optimizing the environment in an agent-based simulation ​
   

• ​A* (A Star) algorithm is a popular path-finding algorithm based off getting to a given destination; decisions are made based off calculated cost. However, this cannot be directly applied because if everyone tries to go to the same exact place the same exact way, crowds and bottlenecking will occur.