Simulations of Emergency Evacuations for Knowledge, Education and Response (SEEKER) is a decision support tool that helps key people in emergency management to make safer and more informed decisions regarding evacuations. This tool is being used in a number of emergency management exercises, for instance, SEEKER was trialled in a recent incident management training exercise involving about 60 emergency management specialists, to plan the response to a mock bushfire evacuation within the Surf Coast Shire, Victoria [1]. SEEKER supports timely decisions by providing information on: 

  • the extent and severity of disaster (e.g., wildfire, flood) impact on the community; 
  • complications associated with large numbers of tourists, major events, and transient populations in the region;
  • expected response of community members to the fire situation and official warnings;
  • impact of activating traffic management plans given available resources;
  • trigger points for decision-making;
  • road speed and capacity constraints with respect to evacuating and background traffic; 
  • unplanned consequences of traffic accidents or blockages as a result of trees over roads; and 
  • evacuation outcomes against a base case of no evacuation.

Modelling adequate representations of human behaviour is an important aspect in developing more realistic evacuation simulations. SEEKER has a sophisticated data-driven human behaviour model that is built upon the BDI-ABM framework developed by RMIT University. This framework allows Belief-Desire-Intention (BDI) cognitive agents to be embedded in an agent-based simulation (ABM). Conceptually, an agent’s “brain” is modelled in the BDI system, while its “body” exists within the ABM system. In an evacuation setting, this infrastructure allows capturing how people make decisions based on their individual goals (e.g., pick-up children from school) as well as factors external to them (e.g., seeing fire/smoke). SEEKER uses the efficient Jill engine (as the BDI system) and MATSim (as the ABM system), a state-of-the-art traffic simulator. Moreover, SEEKER supports generating various kinds of synthetic populations such as census-like populations (consisting of multi-family households constructed based on census data from the Australian Bureau of Statistics), seasonal populations (represents residents and tourists, using data from local emergency personnel [2]) and populations based on bushfire archetypes (recent behavioural research that characterises how people make decisions during evacuation situations).


As the figure shows, SEEKER takes three inputs:

  • Road network:  the key roads of the region.
  • Disaster model:  an estimate of the disaster progression. Users can select a particular bushfire progression layer and set a fire ignition  time:
  • Population: a baseline representation of the population in the region. The video below shows a population of 10,000 people in the Surf Coast Shire region, engaging in their day-to-day activities (Beach, Home, Other, Shops, Work) at different times of the day prior to the evacuation (note that there is no interaction yet between the fire progression and the people, as this is just a visualisation of the inputs). 


SEEKER provides a number of outputs for scrutiny of an evacuation scenario. 

  • Simulation of vehicle movements at a granular level. Below is a simulation of a bushfire evacuation in Maldon, Victoria.
  • Area burnt from the bushfire (Area Burnt), time of impact from fire (Affected times)  and the number of people in each zone (Population in Zones). 

Related Publications