Urban dynamics can be modeled using individual based approach, or more aggregated spatial processes such as diffusion. The model presented here combines geographic and economic approaches : it clearly acknowledge the complexity of urban systems and the role of history in urban dynamics. The agent-based approach adopted, grounded on a simple land-use transport interaction model, tries to embrace the multiple collective decision taken by various actors : metropolitan authorities, local government bodies, etc. New transport infrastructure development is endogeneous, based on rationality of a collective agent, called "planner" in the model. To assess the importance of political fragmentation on urban dynamics patterns, a great variety of planners and administrative divisions can be implemented. Among thematic hypotheses that can be tested with such approach : - to what extend can polycentric metropolis (defined as a non centralized system of governance) enhance local and low distance commuting compared to more centralized ones ? - to what extend political fragmentation can lead to greater car use, caused by a lack of metropolitan-wide transit systems ? ; - under which initial spatial configuration those adverse effects compensate ?