SNCF Transilien is a major operator of Paris suburban trains, which witnessed an increase by 30% of its daily passengers over the last decade. Congestion arising from such a growth has added every year more pressure and more complexity to the operations of the transportation system. In this paper, we present a three-step method to comprehend and to model a whole multimodal mass transit system, including its passengers' behavior. First, we develop a simulation tool to model the multimodal system on a restricted geographic scope, based on complex systems modeling. In parallel, we study the passengers' behavior through an on-site enquiry; we take a particular interest in their behavior while facing disturbances compared to normal situation. Finally, the results of the behavioral approach will be integrated into the simulation prototype using a multi-agent model. The resulting tool will allow Transilien operators to anticipate passenger flows in case of disturbances.