In Case of Emergency...
Social and cultural characteristics in evacuation simulations

Dr Natalie van der Wal
Natalie is a Marie Skłodowska-Curie Research Fellow. She is a member of the Centre for Decision Research (CDR) and the Socio-Technical Centre (STC) at the Business School. Her research expertise lie in agent-based modeling and social science.

Few people take immediate action when the fire alarm goes off. It’s probably a drill, so they finish writing their email, save their files or reach for their belongings. In such a situation, social and cultural (socio-cultural) factors can therefore have a major impact on evacuation time. Current evacuation simulations can become more realistic by basing the models of human behaviour on actual behaviours.


The problem with traditional evacuation models

There are computer simulations that you can use to view and analyse an evacuation of a building, ship, or public space. These computer simulations are often very powerful and can visualise evacuations in 2D or 3D, predicting evacuation times. They can also be useful for planning the layout of events or buildings. Many of the current computer simulations are based on a model that does not simulate social and cultural characteristics of people. In this model, people are like molecules that automatically move back and forth like robots. For example, there are movement rules which dictate the distance people keep from each other and from obstacles. These are based somewhat on social principles of attraction and repulsion. However, in most computer simulations we find behaviours that are just not realistic. For example, often people do not move immediately to the exit as soon as an alarm goes off, nor as soon as the simulation starts. You don’t see people pushing, they do not fall, they do not help each other and they do not get injured.

People can pass on emotions and knowledge to each other, such as fear


Human behaviour

Actual evacuation behaviour is different. Multiple analyses of actual evacuations and behavioural experiments show that people:

Often choose the familiar route out instead of the nearest exit.

Do not immediately evacuate when an alarm goes off, especially when no direct threat is visible.
Can fall, push and show other impatient behaviour.
Do not always understand or follow evacuation instructions.
Can travel in groups, blocking others or even first collecting another group member before evacuating.



If we integrate social and cultural characteristics into emergency programming it will produce more realistic simulations and enable us to more precisely predict the evacuation times and survival probabilities.

My research seeks to improve speed and survival rates in emergency evacuations, through the combined insights gathered from emergency responders, computer modelling and risk communication.

I have analysed footage from past emergency situations and spoken with fire, ambulance and police personnel about their experiences during such incidents. This has helped me to identify which risky behaviours occur in emergency evacuations and has provided me with insights into their underlying causes.

From this I created an evacuation model that contains a number of socio-cultural factors not usually present in traditional simulations. In my model there is a mixture of people both familiar and not familiar with the given environment, and there are those who follow designated evacuation procedures and those who do not. People can pass on emotions and knowledge to each other, such as fear. They can affect each other’s flight route decisions and they can exchange information on whether they believe that there is a threat. People can help each other to stand up if they have fallen. People can travel in a group or alone, speak a certain language and have a certain age and gender that, among other things, influences the speed and chance of behaving in a way that will help others.

In my analyses I focused on human behaviour in ‘bare’ spaces, so spaces with no stairs, ramps or pillars. This is so we can predict the ‘pure’ effect of socio-cultural behaviour of people on evacuation time. The results show that the socio-cultural factors have an effect of three percent delay to an acceleration of up to 30 percent of the evacuation time. However, interestingly, helping each other during an evacuation can accelerate evacuation times. This is partly based on the fact that people can spread knowledge more quickly in groups than when travelling alone. Another explanation is that some social or cultural factors, such as helping others or having different cultural backgrounds, can lead to more phased evacuations – not everyone goes to the exit at the same time, so there are fewer congestions and the entire group evacuates faster.

It will be necessary to test these simulations in different and more complex environments in order to measure the combined effect of socio-cultural and spatial factors, but it is hoped that these early results will drive bigger and better simulations and ultimately save lives.

Find out more

Centre for Decision Research