Abstract: - In the last years unfortunately some tragedies have occurred inside tunnels (i.e. Mont Blanc Road Tunnel in 1999 and others) causing many deaths and serious injuries for several people, so a greater sensibility towards safety, both in terms of infrastructure and devices but also in terms of people awareness, has been grown. The authors propose to study the importance of the human behavior and all the factors involving one’s choices during the evacuation phase inside a long motorway tunnel. In order to take into account all the criteria that affect the choices, especially human factors typical of a PECS model, which involves Physical, Emotional, Cognitive and Social behaviors, a multicriteria analysis has been conducted, using Modeling and Simulation (M&S) reproducing a decision network thanks to Saaty’s AHP (Analytic Hierarchic Process). This model is quite similar to the classical PECS models proposed such as Adam Model, in which Adam, thanks to his attributes and knowledge, tries to survive inside his world exploring it and avoiding all the potential dangerous situations; in fact, the person inside the tunnel, thanks to his/her physical, emotional, cognitive and social attitudes (these ones not present in Adam’s world where he is alone), has to manage how to survive from a fire or a smoke intoxication reaching safety as soon as he can even sacrificing, at least, an expensive good as a car is. Keywords: - Emergency Procedures, Human Behavior, PECS, Simulation, I. INTRODUCTION Safety in tunnels has become important in the last years because a significant price has been paid in terms of human lives inside these structures; in fact, especially after the tragedy inside the Mont Blanc Tunnel in 1999, in which 39 people died and hundreds were intoxicated or seriously injured, the European Union introduced very restrictive regulations about safety devices in road and railway tunnels, especially for those longer than 1000 meters, introducing for instance the emergency tunnel, which is parallel to the normal one and guarantees a safe escape way on foot, smoke detectors, ad hoc traffic lights which break the vehicle flow, especially heavy, in case of accident or high pollution level. Manuscript received November 22nd, 2008: Revised version received December 31, 2008. Authors are from DIPTEM, University of Genoa, Via all’Opera Pia, 15 16145 Genoa, Italy. E. Briano is also from DIP Consortium, New Voltri Terminal, Office Tower, 16158 Genoa GE, Italy. (e-mail: enrico.briano@dipconsortium.org , roberto.revetria@unige.it ). T. C. Author is with the Electrical Engineering Department, University of Colorado, Boulder, CO 80309 USA, on leave from the National Research Institute for Metals, Tsukuba, Japan (e-mail: author@nrim.go.jp). Anyway the people that are involved in an emergency situation, because every people behave and react in different way, provide a great importance also, in order to guarantee the highest percentage of safety, because a wrong choice, maybe guided by the hurry or by fear and anxiety, could be paid at a high price. The aim of this research is in fact to study the people behavior during an evacuation of an about 4,5 km long motorway tunnel, considering all the parameters and the phenomenon which affects the people’s behavior II. STATE OF THE ART ON CROWD BEHAVIOR SIMULATION Crowd behavior simulation has been an active field of research, since the 90s, because of its utility in several applications like emergency plannings or evacuations, it is also useful to study the design and planning of pedestrian areas, the people flow in subway or rail road stations and also in education, training and entertainment. Lee and Son used Believe Desire Intention (BDI) architecture to model a crowd simulation useful for human decision making and planning and using different techniques, like the Bayesian Belief Network (BBN), Decision Field Theory (DFT), and Probabilistic Depth First Search (PDFS), for related sub-models [15]. BDI is a model of human’s reasoning process, where its mental state is characterized by three components: beliefs, desires, and intentions [3]. Beliefs are information which human has about the circumstance, and may be incomplete or incorrect due to the nature of human’s perception. Desires are the states of affairs which human would wish to be brought about. Intentions are desires which human has committed to achieve. In this case they applied the methodology for modeling an evacuation of Washington D.C. because of a terrorist bomb attack: starting from the scenario they characterized different types of agents based on: familiarity with the area, risk taking behavior, confidence index, guidance by police A Study of Crowd Behavior in Emergency Tunnel Procedures Enrico Briano and Roberto Revetria INTERNATIONAL JOURNAL OF MATHEMATICS AND COMPUTERS IN SIMULATION Issue 4, Volume 2, 2008 349