- Increasingly congested surface transportation network in urban areas and grow-ing land values make underground transportation systems more attractive for high-ways (i.e., tunnels) and metro system compared to other options . An underground transportation system can preserve the land above for recreational parks, commer-cial buildings, residential homes, or other purposes while providing an eﬃcient, cost-eﬀective underground corridor to move people and goods by separating from the sur-face system . However, the underground transportation systems present safety and operational challenges as well if incidents (e.g., ﬁre, ﬂood, terrorist attacks) occur. Major tunnel incidents since 1995 have killed 713 people worldwide . From 1999 to 2001, several tunnel ﬁres with multiple deaths occurred in Europe. For example, 39 people died in the ﬁre in the Mont Blanc Tunnel between France and Italy in March 1999, 12 people died in the ﬁre in the Tauern Tunnel in Austria in May 1999, and 11 people died in the ﬁre in the Gotthard Tunnel in Switzerland in October 2001 in which the temperature reached 1,000 degrees Celsius (°C) (1,832 degrees Fahrenheit (°F)) within a few minutes . These incidents caused signiﬁcant safety concerns regarding underground transportation system safety. This problem is complex for multiple reasons: (1) how people will react in tunnel emergencies is unpredictable, (2) ﬁxed entrances and exits, (3) evacuation is likely to be self-initiated, (4) high-density presence of pedestrians, and (5) diﬃcult to access for ﬁrst responders and emergency vehicles.
The objective of this thesis is to present an interdisciplinary agent-based evac-uation modeling framework for emergencies in underground transportation systems. Through this established framework, we will identify and validate the critical factors which aﬀect life safety in underground emergency scenarios. The identiﬁcation of the critical factors is validated by empirical data from historic underground tunnel ac-cidents. The evacuation model is built through an agent-based platform: Anylogic. Then, a multi-discipline framework is introduced to analyse and identify problems related to evacuation in underground transportation systems. Finally, we study in detail and simulate the eﬀects of ticket gate type, walking speed, gender, group size, pedestrian’s density, and smoke on evacuation time. The research results from this thesis will provide decision-making support and guidance for government decision-makers, design engineers, and agency professionals to optimize underground station design. The experiment results indicate that the proposed agent-based underground transportation emergency evacuation modeling framework in this thesis is eﬀective at evaluating the impacts of the identiﬁed critical factors on evacuation eﬃciency and life safety.