View non-flash version
will evacuate via the main escape routes unless otherwise stated; and full availabil -ity of escape arrangements is considered. In addition, a safety factor having a value of 1.25 is introduced in the calcula -tion to take account of modeling omissions and assumptions. ese include the follow -ing: that the crew will immediately be at the evacuation duty stations ready to assist the passengers; that passengers follow the sig -nage system and crew instructions (i.e., route selection is not predicted by the analy -sis); that smoke, heat, and toxic re products present in re euent are not considered to impact passenger/crew performance; that family group behavior is not considered in the analysis; and that ship motion, heel, and trim are not considered. e requisite scenarios (benchmarks) are dened as follows: r þÿ B O J H I U D B T F T D F O B S J P X I F S F B M M Q B T T F O H F S T are in their cabins r þÿ B E B Z D B T F T D F O B S J P X I F S F B M M Q B T T F O H F S T are in public spaces r þÿ U X P P U I F S T D F O B S J P T P O F G P S E B Z T D F O B S J P and one for night scenario with reduced escape route availability. e evacuation time is calculated and then compared with the performance standards, which are: Evacuation models Evacuation simulation models can be divided into two main categories: flow models and agent-based models. The first are at a mac -roscopic level, where people movements are modeled as flow going through ?pipes? in a graphical representation of the geometry. e second type of model is at a microscopic level where persons are modeled as individuals inter -acting among themselves and the environment; hence, the geometry is modeled in detail. ese models are called multi-agent models. e main advantage of the rst type of model is that it is quite easy to implement and make use of network ow algorithms and methods. However, it fails to capture important aspects as it does not model, for example, the interaction between agents like avoidance, group eect (where people tend to move as a group at the pace of the slowest member), or the counter-ow eect when people move in opposite directions, aspects for which the second type of model accounts. Another category of model is the meso -scopic model, which combines both micro and macro levels of planning. At the micro level, rules are in place to govern agent movements from way nding to avoidance strategies, while at a macro level, there is the path planning process. Models can be further distinguished by their modeling of the environment (geometry) where some models are grid-based while others model the environ -ment as a continuum. Agent characteristics, such as speed, age, gender, and reaction time (awareness) are specied using prob -ability distributions. Evacuability index: EVi At the Ship Stability Research Centre of the University of Strathclyde, an agent-based evacuation simulation tool specifically tailored to the ship environment has been developed over the past few years, using SafetyatSea Ltd.?s EVi pedestrian movement simulation software package. Following is a brief description of its modeling. Multi-agent modeling An agent is dened as an encapsulation of code and data, capable of executing inde -pendently the appropriate piece of code depending on its own state (the encap -sulated data), the observables (the environment), and the stimuli (messages from other parts of the system or interac -tively provided). e agent?s action model is a sense-decide-act loop. July 2012 www.sname.org/sname/mt 1.25(A + T) + (E + L) n(E + L) 30 min WHERE :A :rep resents awareness time T :rep resents travel time E :re presents embarkation time for the life saving appliances L :re presents launching time of the life saving appliances n :þÿ represents 60 minutes for ro-ro passenger ships and for passenger ships with less than 3 main vertical zones (MVZ), and 80 minutes for ships with more than 3 MVZs. Agent characteristics, such as speed, age, gender, and reaction time (awareness) are specied using probability distributions.