TOWARDS AN HYBRID SIMULATION APPROACH OF TRANSPORTATION SYSTEMS Mohamed Said EL HMAM, Hassane ABOUAISSA, Daniel JOLLY, Amar BENASSER Laboratoire de Génie Informatique et d’Automatique d’Artois Faculté des Sciences Appliquée, Technoparc-Futura, Béthune, FRANCE {mohamed.said.el.hmam, daniel.jolly}@fsa.univ-artois.fr, {hassane.abouaissa, amar.benasser}@ univ-artois.fr Abstract: The hybrid model presented in this paper is a mixture of two models: macroscopic and microscopic. This approach presents two major advantages. The first one is the easiness of treatment for the wide-area networks provided by the macroscopic model. Secondly, the good results provided by the microscopic model when it is a question of solving the problems related directly to the vehicles. Such model is efficient if and only if the transition between models is guaranteed. This task is difficult to realise and requires more attention and some conditions have to be checked. This paper proposes a reduced hybrid model to be implemented and validated in the further work. Copyright © 2004 IFAC Keywords: Hybrid model, macroscopic model, microscopic model, Simulation, Transportation Systems. 1. INTRODUCTION The systems simulation activity has been considered for several years the only means allowing to apprehend their complexity and to better understand their internal behavior. This activity naturally found its application in the field of the transportation systems, to model the traffic phenomena and, to elaborate effective strategies and algorithm for their regulation and control. However, simulate a system in general, requires disposing of an adequate model. In this context, several modeling and studies approaches of the transportation systems and traffic in particular have been proposed. They can be classified into two mains categories. The first one, the macroscopic (Messmer et Papageorgiou, 1990) approach, considers the traffic from a continuous point of view and it is composed of a set of vehicles in flow in various sections of the road. Three basics variables are used to represent the traffic flow. The density (or concentration) of vehicles K , the flow Q and the mean speed V . The macroscopic models are very widespread because they allow an easiness of treatment for the wide-area networks. In addition, they provide a good basis to design effective traffic flow control strategies in order to avoid or to minimize the congestion phenomena. Nevertheless, macroscopic models and simulators present some limits that consist on their incapacity to take into account the vehicles interactions such as conflicts between vehicles, discontinuity in the flow introduced by the traffic signals and non-homogeneity between adjacent lanes (Leclercq et Ludovic, 2002). Indeed, such information of the inter-vehicles conflicts is necessary to understand the exact causes of congestion phenomena. For this, the second approach, based on microscopic considerations allows to mitigate this insufficiency by offering real representation possibilities of the interactions and conflicts between vehicles. In this context, several microscopic models have been proposed. Among these, the “Car Following Models” (Gabard, 1991) represents the most adequate one. Indeed, the “Car-Following Model” laws aim to translate the behavior of the vehicle driver in a flow IFAC DECOM-TT 2004 Automatic Systems for Building the Infrastructure in Developing Countries October 3 - 5, 2004 Bansko, Bulgaria