Modeling and simulating activities in and around train stations is a challenging task. Indeed, it implies the consideration of a dynamic and heterogeneous set of actors, obeying to different rules and control systems. In this paper, we propose a multi-agent model for such simulations. The proposal comprises three parts. The first part is a multi-agent simulation model of travelers together with their interaction with the station platform. Passengers adapt their behavior to the other travelers while pursuing their travel objective, and in interaction with the available information sources, either local at the station, or personalized (through their smart phones for instance). The modeling of this traveler's information system is the second part of the model, where transportation operators are able to define strategies regarding the type of information to be provided, their frequency, and the medium used to broadcast it. Finally, the third part of the model is the management platform of the train station, which monitors both trains and pedestrians behaviors in the station. The platform is responsible for the optimization of the traffic fluidity and passengers' safety and proposes actions related to crowd deviation and access control. The architecture advocates for a closed-loop modeling between the simulation, the train station management and the traveler information, allowing for the testing of whole control systems instead of specific strategies.