​

Executive Summary 
 

D5.3 is a demonstrator deliverable that comprises novel crowd simulation algorithms based on incident propagation forecasting models, and is known as the LDER component (Location Based Dynamic Evacuation Route). The software and algorithms being developed are not easily accessed and usable by a non-expert. Therefore, this report describes relevant background research, existing state of the art and high level description of the novel algorithms required to provide real-time routing during evacuation in the  SafePASS project.

 

The LDER component is made up of many different system parts that can be separated into 3 main parts, all of which are being developed from a base existing state of the art that is to be progressed to allow for the complexity of large passenger vessels to be modelled and managed in real-time:

​

• LDER route calculation, which calculates the optimum routes

• SafePASS crowd Model Designer (SMD), which is a visual software tool used to input the ship layouts, routes and simulation parameters to be used in the route calculation

• Fire and flood propagation simulations, which model the spread of fire, smoke and flood in the ship

• These algorithms will be developed to model crowd movement in any scenario encountered on ships, such as unavailability of escape routes; ship motion; behavioural factors such as family groups wanting to be together; and passenger demographics that affect movement of passengers, such as age.

​

The current regulations and guidelines are clear for implementation on ship design and for enacting evacuation procedures, but do not yet take into account the presence of LDER calculations that would allow many different evacuation strategies. These include dynamically allocating muster stations, directing passengers directly to embarkment of lifeboats, lateral evacuation across vertical zones of the vessel or evacuation of a ship in port. These can be modelled in real-time and the LDER calculations will provide the requisite routes for passengers to follow depending on the chosen evacuation process.

​

Working much faster than real-time, the LDER will take data processed by the SafePASS core engine such as sensor data and estimated counts of passenger location to run a live simulation during the evacuation and feedback results on congestion, evacuation times for each component of evacuation and enable a much better situational awareness and guiding of passengers.

​

A validation plan has been created that includes key performance indicators in which the algorithms can test against as they develop. This includes Key Performance Indicators (KPIs) and identification of data sets that crowd models can be validated against.

​

The evacuation strategies will be affected by future results of the work performed in other deliverables and from End User feedback. Therefore, the initial framework presented in this deliverable will further be evolved during the project’s lifetime.