Simulation is a means to help urban planners and investors to optimize inhabitant satisfaction and return on investment. An example is the optimal match between household preferences and property profiles. The problem is that not enough knowledge exists yet about dynamic user activity models to build reliable and realistic simulators. Therefore, we propose a modelling and software technique that produces simulator prototypes very efficiently for the development, test, and evaluation of many different user activity models, using executable models, code generation, and a domain specific software process. As a specific feature, the model is based on many agents acting independently from each other and that are mapped in several refinement steps into the same number of concurrent processes. The housing example is used as a case study to explain the process and show performance results.