The need to save energy at the urban scale leads to study how numerical simulation and optimization methods can help the architects to design buildings and districts having the best possible energetic performances, regarding daylight, warming or cooling, and photovoltaic capacities.The present work uses a physical model of irradiance and an evolutionary algorithm to optimize the repartition of houses inside a fixed area, in order to improve the total solar radiation received.Several objectives can be considered in order to save energy: during cold periods or in cold regions, a maximum of solar energy must be received by the buildings, but we could also try to minimize energy consumption for air-conditioning, that means to reduce the solar exposure in hot periods, or we could formulate a multicriteria problem.As a first test, we choose to maximize the irradiance function evaluated during the shortest day of the year, taking into account the influence of the surrounding buildings.