In the planning and evaluation stages of infrastructure projects, it is necessary to manage huge quantities of information. Cities are very complex systems, which need to be modeled when an intervention is required. Such models allow us to measure the impact of infrastructure changes, simulating hypothetic scenarios and evaluating results. This paper describes a methodology for the automatic generation of urban space models from GIS sources. A Voronoi diagram is used to partition large urban regions and subsequently define zones of interest. Finally, some examples of application models are presented, one used for microsimulation of traffic and another for air pollution simulation.