In multi-agent applications, normative systems are usually used to regulate the behavior of the agents. They provide an efficient means to ensure limited deviations from an expected ideal behavior. Many works have been done in this classical research direction, less frequent are the works on norms in simulation. In this paper we focus on simulations of spatially situated agents, typically moving around simulated physical environments. Our goal is to provide a mechanism to generate efficiently consistent agents' characteristics. We propose to model behavioral differentiation in such simulations as a violation of the norm, and show its application to traffic simulation with the driving simulation software used at Renault, scaner c ii.