Padang Article Swipe
YOU?
·
· 2016
· Open Access
·
· DOI: https://doi.org/10.5334/baw.76
· OA: W4237754167
The Padang scenario demonstrates the MATSim application to large-scale evacuation problems.The scenario has been created as part of the third party funded project "Last-Mile".Taubenböck et al. (2013) give a comprehensive overview.Padang is located on the west coast of Sumatra Island, Indonesia.In 2014, the city had a population of about 1 000 000 people.Because of its problematic location on the coast in a so-called "seismically locked" area (McCloskey et al., 2010), Padang is prone to earthquakes and subsequent tsunamis.In the "Last-Mile" project, a realistic tsunami scenario, triggered by an earthquake about 300 km o the coast, was identi ed (Goseberg and Schlurmann, 2009).The assumed tsunami would leave about 30 minutes for the evacuation.The ooding would reach as far as three kilometers inland, thus threatening up to 330 000 lives.Lämmel (2011) developed a MATSim scenario representing the city with its a ected population.One unusual aspect of the Padang situation is the expected universal evacuation by foot; simulating pedestrians with MATSim was a novelty when this project started.The standard simulation model (see, e.g., Section 1.3) was thus adapted to deal with pedestrians.Details are discussed by Lämmel et al. (2009).Another important variation, contrary to most standard transport scenarios, is that network links would ood once the tsunami reached them.Thus, accessibility-ooded or not ooded-of the network links is time-dependent, which is modeled by a time-dependent network (Lämmel et al., 2010).In the time-dependent network concept, link attributes-like freespeedcan be changed, while the simulation is running, by precomputed network change events.For the Padang scenario, the network change events have been extracted from microscopic ooding simulation data.Key Padang scenario facts:• The network consists of about 6 000 nodes and 17 000 links.• Synthetic populations for morning, a ernoon, and night have been created, containing up to 330 000 agents.
