Agent-Based Modelling of Tsunami Evacuation Accessibility in Coastal Cilacap

Abstract

Coastal areas adjacent to active subduction zones are at risk of tsunami disasters. Therefore, it is crucial to develop evacuation plans to minimise casualties should a disaster occur. This study examines evacuation accessibility along the coast of Cilacap City, Indonesia, using agent-based modelling. The modelling was conducted using the MATSim platform, which integrates the road network, the distribution of buildings as evacuation origin points, and population distribution data. The simulation involved 4,979 individuals from three coastal villages, using various modes of transportation, such as walking, motorcycles, and cars. The simulation results show that each mode of transportation has a different travel time for all agents to reach the evacuation point. Evacuation on foot took 26 minutes for all agents to reach the evacuation site, 13 minutes by motorcycle, and 4 minutes by car. The effective evacuation time (golden time) was set at approximately 35 minutes; within this timeframe, 90.1% of agents successfully reached the evacuation site, while 9.9% exceeded the golden time and were at risk of being affected. These results highlight the importance of integrating spatial data with an agent-based approach to identify weaknesses in evacuation accessibility and support evacuation planning for vulnerable coastal communities.