Aircraft Evacuation DEVS Implementation & Visualization Patrick Castonguay & Gabriel Wainer Department of Systems and Computer Engineering Carleton University 1125 Colonel By Drive Ottawa, ON K1S 5B6 pcaston3@connect.carleton.ca ; gwainer@sce.carleton.ca Keywords: Cell-DEVS, CD++, Blender, Visualization Abstract Emergency evacuation from new, higher than normal, aircraft is believed to be affected by a new variable: passenger’s hesitation at the exit door. This paper discusses how a previous simulation studying this effect was replicated in Cell-DEVS using the CD++ toolkit and how the results were visualized using the Blender Python interface. We will show how one can relatively easily apply advanced visualization techniques to any DEVS simulator results. As well, we noticed that the initial design of the model and simulation can limit the level of visualization which can be achieved if that design is not intended for visualization in the first place. 1. INTRODUCTION With the advent of new twin-decks Very Large Transport Aircraft (VLTA), concerns about the speed at which they may be evacuated have emerged [1]. Many factors come into play when conducting an aircraft emergency evacuation [2, 3, 4]. This paper focuses on the possible effects passengers hesitation could have on total evacuation time when exiting from an abnormally high elevation [5]. Amos and Wood developed and applied a model of the emergency exit procedures from the Airbus A380 to study the effect of individual delays at the emergency door when exiting a plane [6]. Here, we will show the definition of these models using the Cell-DEVS methodology [7]. Cell-DEVS is an an extension to DEVS [8] (Discrete Event Systems Specifications) which enables efficient execution of cellular models. The approach extends traditional Cellular models defining each cell as a DEVS atomic model and the space as a DEVS coupled model, including a flexible way of defining the timing of each cell. The CD++ toolkit [9] was used to implement the model and generate the initial results. Then, the Blender toolkit [10] was used to integrate the simulation results into a 3D visualization environment, based on previous experiences in this area [11]. Results from DEVS simulation can be visualized in many different ways. The CD++ toolkit provides DEVSView as well as the CD++ Modeler add-on. These tools provide a quick representation of the model and simulation behavior in the 2D space. 3D visualization of CD++ results have previously been done using VRML, Atlas, Maya and Blender [12, 13, 14]. We extended previous results in order to demonstrate the flexibility and portability of both the CD++ tool kit as well as Blender. 2. BACKGROUND Emergency evacuations of any aircraft have to meet certain regulations; all passengers and crew have to be evacuated within 90 seconds, with only half of the emergency exits available. In [6], it was argued that, due to the dangerous and expansive nature of the evacuation, real demonstrations are normally limited to a single instance. As well, for commercial reasons, the results are normally kept secret within the company. The effect of the delay at the door was suspected to be of importance due to the nature of the second deck of the A380 which was much higher than other aircraft at the time. The model was initially implemented using NetLogo programming language [6]. Figure 1 represents the floor layout of the A380 which was used as a construct for the original work. Figure 1: A380 upper floor plan We implemented this agent based cellular automata using Cell-DEVS [7]. A cell grid representing the floor plan is defined. A single cell represents a wall, a passenger, an exit or an empty space. Each cell contains information relative to the nearest exit (id & distance). Each agent (passenger) has the following attributes: p i : grid position; h i : heading; m i : max speed s i : current spead; and d i : delay at door.