Power, W.L.; Lukovic, B.; Wang, X. 2021 Agent-based tsunami evacuation modelling of the Wellington CBD. Lower Hutt, N.Z.: GNS Science. GNS Science report 2021/04. 25 p.; doi: 10.21420/XPRG-9529
Abstract:
The Wellington CBD is at risk from tsunami, which in some cases could arrive as soon as 30 minutes after an earthquake. The primary means for saving lives is to educate the public to self-evacuate from tsunami-prone areas if they feel a long or strong earthquake: “Long or strong, get gone!”. In densely populated areas like the Wellington CBD, Civil Defence and Emergency Management (CDEM) encourage people to evacuate on foot if they can. Mass evacuations pose many problems in urban areas, as typically the urban infrastructure is not designed to handle such large volumes of pedestrians. In rural areas, these problems can be identified using ‘tsunami evacuation hīkoi’ exercises, where the community evacuates as if a large earthquake has occurred. This is not practical in a large city like Wellington with tens of thousands of residents and workers. Instead, we have used a computer model of people evacuating on foot from the CBD and have used this to try to identify some of the problems that are likely to occur and their potential solutions. Our initial model identified a problem with the narrow passageways between Lambton Quay and The Terrace – for many people in the CBD these represent the shortest, and theoretically quickest, route out of the tsunami evacuation zone. However, these narrow passageways were not designed for such large volumes of foot traffic and in our simulations rapidly become choked with people trying to use them. A revised model was developed in which (a) only a proportion of the population was aware of these passageways, and (b) those who did attempt to use these passageways would change to an alternative ‘Plan B’ of using only the major roads to evacuate if they were stuck for too long while attempting to enter the passageways. With this revision in place we find that in daytime approximately 65,000 agents (simulated people) are able to reach safety within 30 minutes of a simulated earthquake out of about 85,000 people present in the evacuation zone. It is important to stress that this scenario assumes that the evacuation takes place within an undamaged environment. At night there are far fewer people present in the CBD, and most of the approximately 9,000 are able to evacuate in time. For both day and night, scenarios estimates were made of the numbers of people who might seek safety in different parts of town outside the evacuation zones. A selection of variations to our standard scenario looked at the effects of changing the proportions of people attempting to use the passageways, the effects of assuming greater time to leave office buildings, the effects of modifying one of the passageways to accommodate more people, and an approximation of the effects of earthquake damage which was assumed to lead people to avoid walking within 4 m of buildings to avoid falling glass and other debris. Some key conclusions are: • It is best if most CBD daytime occupants evacuate using only the roads, as the narrow passageways between Lambton Quay and The Terrace cannot cope with large numbers. Even with modifications to passageways to accommodate more people, it may still be problematic if too many people attempt to use them. • It may be preferable to use signage to encourage people evacuating from the Civic Square area to evacuate along Victoria Street rather than on Willis Street, to avoid overcrowding. • Our simple model of the effects of building damage reduced the estimated number of people reaching safety in 30 minutes by about 10,000. Further research to develop a more accurate model of the effects of damage is recommended. This will also help to guide advice on the optimum proportion of people who could be advised to remain in the CBD by using ‘vertical evacuation’ within tall earthquake- and tsunami-resistant building. (auth)
