Modeling School-Bus-Shinkansen Evacuation Strategies for Volcanic Disasters: A Case Study in Kagoshima, Japan

This study examines microscopic evacuation simulations using predefined bus fleets to move residents from affected areas to safety, with a focus on volcanic disasters, an often-overlooked Scenario compared to earthquakes or floods. Stratovolcano eruptions, such as those from Sakurajima near Kagoshima City, can trigger prolonged crises, with massive ash and pumice severely disrupting roads and urban infrastructure. In dense cities, private vehicle evacuations risk extreme congestion, making organized, large-scale bus-based strategies essential. This research models bus fleet operations under conditions of an imminent eruption. Besides following the suggestion of the City of Kagoshima for each household to use the passenger car for evacuating to respective shelters outside the city, we reason on how to evacuate the most vulnerable road users, namely children and elderly people in a fast and safe way, especially during typical working hours of a weekday in the central business district of the city. Therefore, we calculate bus routes connecting specified building types (in our example schools) with the Kagoshima-Chuo Shinkansen Station, enabling the fast evacuation (e.g., to Kumamoto) of a large number of persons. For reasoning about optimal setups for this strategy, we define three different scenarios that use the same traffic flow simulation network, but varying road user agents compositions with (Scenario 1) only simulating evacuation buses, students and teachers entering and exiting them (and walk to a train platform), Shinkansen used to transport the evacuees to Kumamoto Station, (Scenario 2) addition of passenger-car demand estimated from real traffic counts at 287 map-matched detectors on August 5, 2025 (for the Japanese standard time between 7:15 and 13:15), and (Scenario 3) all students and teachers are walking (as pedestrian road users) from respective school bus stops to the Kagoshima-Chuo Shinkansen Station (with passenger-car traffic demand from Scenario 2). After discussing the respective simulation outcomes, we suggest further simulation scenarios for future research.