Since the Black Saturday fires in Victoria, Australian emergency officials have emphasised evacuation planning and preparation, but such planning can fail to consider and account for multiple evacuation scenarios. This includes ‘credible “worst case” scenarios’ which ‘stress current capabilities’.1
Worst case scenarios are those that pose the biggest threat, strain emergency response operations, and potentially endanger people. For example, rapidly spreading fire conditions can block roads originally allocated as evacuation routes and leave residents trapped.2 Fires can also cause power losses,3 hindering the receipt and implementation of evacuation orders. In such cases, evolving fire conditions affect evacuations. This should be considered when planning for bushfire evacuation – especially when assessing the robustness and resourcing of a plan and developing contingency plans.
Little guidance exists on the scenarios that decision-makers should consider when planning for bushfire evacuation, making it difficult up-front to know which will be most credible. Evacuation modelling projections can help to identify the scenarios that pose the greatest challenges.
The Royal Commission into National Natural Disaster Arrangements mentioned the need to model events using credible scenarios and projections,1 considering climate, weather, hazard, fire behaviour and other contributing factors.
However, while modelling a wildfire’s spread and severity is critical and has enormous benefits, an understanding of the vulnerability of a population and its capacity to reach safety requires a quantitative assessment of both the hazard and community evacuation.
Learning from fire safety engineering
Fire safety engineering can help. Engineers determine the level of safety provided by particular building designs as part of a performance-based design (PBD) approach. This involves comparing the required safe egress time (RSET) (the time to evacuate all building occupants) and the available safe egress time (ASET) (how long until conditions become untenable). A building is considered safe where ASET is greater than RSET by some margin.
Fire safety engineers develop credible fire and occupant/egress scenarios for their building design – selected according to their probability and the challenges posed – and use these to quantify ASET and RSET.
This PBD approach could equally be adopted to assess the threats posed to a community by credible bushfire scenarios and planned responses.4 The authors contributed to the development of such a concept for bushfire (WildfireRSET and WildfireASET) as part of a recent international research project funded by the US government, influenced by the original work of Li et al..5

Performance levers
Some guidance exists for fire safety engineers on the types of fire and evacuation scenarios that should be considered when running PBD analyses of buildings.6,7 Nilsson and Fahy7 have suggested that occupant-based evacuation scenarios consider:
- the building and its uses, to account for the types of people who will use that building. This is particularly important when considering a building with multiple functions and, in turn, different occupant populations;
- the layout of the building, since it shows the egress paths available (and their capacity to accommodate an evacuating population). Some exits and paths may be more familiar to occupants than others, which is important to take note of during analysis; and
- the occupants of the building and their response characteristics: permanent vs transitory occupants; trained vs untrained; adult vs young/old; the presence of cognitive, sensory or mobility impairments; awake vs asleep (or unconscious/inebriated); role; and social relatedness.
Scenarios are then formed from building design, use, population characteristics and response factors.
Nilsson and Fahy7 note that these characteristics are important because of the way they can influence evacuation. For example, permanent or trained individuals are more likely to be familiar with the building’s evacuation plan and procedures, whereas others may not. Age and even some types of disabilities may impact a person’s ability to make independent decisions and/or affect their ability to physically move to safety. Other impairments (e.g. sensory) or states (e.g. sleeping) may hinder a person’s ability to receive cues from their environment, signalling that a fire is threatening their safety. Social groupings can play a role since families, for example, are more likely to assemble and move together, and so on.
Nilsson and Fahy7 note that scenario development is an iterative process that requires consideration of the building’s layout, its use, the users of the building, and the fire. The ultimate goal of these analyses is for the engineer to develop scenarios that are severe, yet credible, to test the building’s fire safety design.
How is this relevant to bushfire evacuation?
Such guidance might be adapted to develop evacuation scenarios for communities exposed to bushfire. Here, planners can view the community as an engineer would the building by: (1) identifying the evacuation routes, possible modes of transport and ultimate destinations/locations that provide safety; (2) identifying the community members present, e.g. given building types, and their characteristics (resident vs visitor, trained vs untrained, etc.); and (3) establish the nature of these people’s responses.
These factors can be used to develop credible evacuation scenarios that can then be compared with credible fire scenarios to establish: (1) whether the community can evacuate to a place of safety before fire conditions become untenable; and if not, (2) what types of interventions or design changes (if possible) should be made to allow for safe evacuation. This type of analysis allows evacuation to be quantified and the impact of different factors to be determined.
Many of the occupant characteristics identified by Nilsson and Fahy7 for building fires are relevant to the development of evacuation scenarios for bushfire. That said, there are important characteristics unique to households’ response to bushfire, not necessarily relevant in buildings. Those of lower socio-economic status, for example, may not have access to personal vehicles and may require carpooling or public transport to evacuate. Additionally, individuals with pets or livestock may be less likely to evacuate their homes, especially if shelters are not equipped to accommodate them. Many articles have been written about the factors that influence evacuation decision-making during bushfires (e.g. 8,9), and while less has been written about the factors that influence evacuation movement during bushfire,10 these articles provide some understanding of the characteristics that matter.

Candidate scenarios
Below is a list of several factors (and example settings) that may form candidate scenarios when planning for bushfire evacuation. In each case, a factor is named and then possible settings identified.
Evacuee (resident/community member) response:
- Evacuees (some %) decide to stay and require rescue
- Evacuees (some %) leave 30 mins, 1 hour, 2 hours after the emergency warning is given (These evacuees may be unaware of the fire, unaware of the procedure, etc.)
- Most evacuees (e.g. greater than 60%) start their evacuation movement around the same time (i.e. potential for congestion to occur at particular locations)
Mode choice:
- Evacuees (some %) require public transport or assistance to evacuate
- Evacuees (some %) evacuate with multiple personal vehicles (per household) of various sizes (cars, trucks, campers, trailers, etc.)
Route choice:
- Evacuees (some %) use the most familiar route (e.g. main highway) to evacuate the community, rather than the planned route
- Evacuees (some %) use the longest route to evacuate the community, rather than the planned route (e.g., tourists/visitors)
- Evacuees (some %) choose to travel to other locations within the areas of risk before evacuating (e.g., to pick-up loved ones)
Destination choice:
- Evacuees (some % larger than available capacity) decide to travel to the community fire refuges for safety
Fire conditions:
- After time (X mins), at least one evacuation route becomes blocked due to fire
This last factor might be represented using results from modelling of the fire itself. However, the simple approach identified here might enable the modeller to determine the sensitivity of the community to fire should the fire block various routes before evacuation starts.
Simple example
Figure 1 shows a simple example of how the PBD approach might be applied to the evacuation analysis of a bushfire-threatened community. This community is to have the surrounding road network upgraded, with several designs being considered. This expensive endeavour has to be costed according to traffic demand of the road network variants during routine (non-emergency) scenarios. This upgrade is also examined for its potential impact on evacuation using the PBD approach. The community’s characteristics (i.e. demographics, training/familiarity, longevity, vehicle ownership, etc.), estimated response, viable emergency procedures, and refuge locations are identified and used to develop evacuation scenarios, which can then be simulated to calculate evacuation times. These are then compared against projections of credible fire conditions and how they progress over time.

Analytically, a planner might quickly be able to determine the capacity of the road network designs and intuitively establish realistic traffic loads. However, this will not necessarily capture the dynamics of the evacuation in conjunction with the evolving fire conditions, such as when evacuating populations are most exposed, or where the system is under/over-utilised. In contrast, the use of a PBD will calculate evacuation times and the number of residents not able to evacuate before fire conditions become untenable (for each scenario or across all of the scenarios examined). It would at least provide some basic evidence for comparison – both of the outcomes and the conditions that are produced during the evacuation simulation.
The same community layouts can be tested with different populations (larger numbers, older, less familiar, etc.) to see how susceptible the community is to changing demographics, as well as different fire conditions. Similarly, road availability, resident evacuation delays, vehicle availability, etc., along with fire severity and location can be varied. Such analysis enumerates the vulnerability of the community to bushfire; and provides insights into their capacity to cope with the bushfire scenarios posed and the robustness of the road network to cope with the demand.
Discussion
Without quantifying evacuation under different fire scenarios, it is difficult to develop evidence-based contingency plans to address issues ahead of a fire event. By quantifying the outcome of each scenario (e.g. the time for the community to reach a place of safety) it allows for:
- a comparison between different fire scenarios to determine which incident poses a particular threat
- a comparison between different behaviours or official interventions (e.g. procedural changes, different resource allocation, etc.)
- the identification of underlying conditions that contribute to the projected outcomes (e.g. where routes are overloaded or causing congestion, where routes are underutilised, etc.), suggesting possible changes.
Such an approach has enormous value in routine times where a bushfire might be the only threat posed in a particular area. However, 2020 has taught us the importance of considering simultaneous threats when planning for evacuation. At present, evacuation planners are also forced to consider how the Covid-19 pandemic may influence emergency response procedures and protocols for bushfire events. The Royal Commission report notes that compound disaster scenarios require attention.1
The presence of the pandemic may affect the community response to a bushfire in several ways:11
- a higher or lower number of people may reside in a community because of Covid-19 travel restrictions
- delays or non-evacuation behaviour may occur because individuals are concerned about spreading or catching the virus
- fewer people (than previously planned for) will be able to use public transport or shelters because of social distancing requirements
- evacuees may choose different destinations than previously planned
- emergency responders could fall ill, or be reallocated in the regional/national response to the pandemic, or they may be unable or hesitant to perform some tasks, given the pandemic (e.g. going door to door to notify households of evacuation warnings).
We do not know precisely how such a compound event will affect evacuation response. However, we can identify the underlying factors that it might affect. We might then vary the extent of this effect during evacuation modelling to quantify the potential influence that such a compound incident might have on a bushfire evacuation and help identify the preparations required to mitigate these effects.
For more information, email erica.kuligowski@rmit.edu.au
References
- Final Report of the 2020 Royal Commission into National Natural Disaster Arrangements, https://naturaldisaster.royalcommission.gov.au/.
- Gramenz, J. ‘Major roads closed as thousands flee NSW and Victoria’, Accessed 12 Feb 2021, https://www.news.com.au/technology/environment/major-roads-closed-as-thousands-flee-nsw-and-victoria/news-story/5884afe32359429af35f80b3498f7b6c
- MacDonald-Smith, A. ‘Week-long blackouts feared as power lines destroyed’, Accessed 12 Feb 2021, https://www.afr.com/companies/energy/week-long-blackouts-feared-as-power-lines-destroyed-20200102-p53oc2
- E. Ronchi, S.M.V. Gwynne, G. Rein, R. Wadhwani, P. Intini, A. Bergstedt, e- Sanctuary, Open Multi-Physics Framework for Modelling Wildfire Urban Evacuation, Fire Protection Research Foundation, Quincy, MA, 2017.
- D. Li, T.J. Cova, P.E. Dennison, ‘Setting wildfire evacuation triggers by coupling fire and traffic simulation models: a spatiotemporal GIS approach’, Fire Technol. 55 (2019) 617–642, https://doi.org/10.1007/s10694-018-0771-6.
- G.V. Hadjisophocleus, Mehaffey, J.R., 2016. Fire Scenarios. In: Hurley, M.J., Gottuk, D.T., Hall, J.R., Harada, K., Kuligowski, E.D., Puchovsky, M., Torero, J.L., Watts, J.M., Wieczorek, C.J. (Eds.), SFPE Handbook of Fire Protection Engineering, Springer, New York, pp. 1262–1288.
- D. Nilsson, Fahy, R., 2016. ‘Selecting Scenarios for Deterministic Fire Safety Engineering Analysis: Life Safety for Occupants’. In: Hurley, M.J., Gottuk, D.T., Hall, J.R., Harada, K., Kuligowski, E.D., Puchovsky, M., Torero, J.L., Watts, J.M., Wieczorek, C.J. (Eds.), SFPE Handbook of Fire Protection Engineering, Springer, New York, pp. 2047-2069.
- J. McLennan, B. Ryan, C. Bearman, K. Toh, ‘Should we leave now? Behavioral factors in evacuation under wildfire threat’, Fire Technol. 55 (2019) 487–516, https://doi.org/10.1007/s10694-018-0753-8.
- L.H. Folk, E.D. Kuligowski, S.M.V. Gwynne, J.A. Gales, ‘A provisional conceptual model of human behavior in response to wildland-urban interface fires’, Fire Technol. 55 (2019) 1619–1647, https://doi.org/10.1007/s10694-019-00821-z.
- Kuligowski, E.D., 2020. ‘Evacuation Decision-making and Behavior in Wildfires: Past research, current challenges and a future research agenda’. Fire Saf. J.
- E.D. Kuligowski, Gwynne, S.M.V. ‘Considerations for Planning Community Evacuation During a Pandemic: A Focus on Human Behavior During Wildfire Emergencies’. Fire Protection Engineering Magazine, FPE extra, Issue 53, May 2020. https://www.sfpe.org/publications/magazine/fpeextra/fpeextra2020/fpeextraissue53

