Réseaux complexes et trafic routier ; Road network structure and traffic pattern

The personal automobile has changed the world we live in. In particular, transportation infrastructure has been modified and expanded at a rapid pace during this century to accommodate the growing number of mo-torists. Very quickly, the increase in the number of vehicles made it necessary to establish a high-way code, to make traffic flow more smoothly and safely, but also to study this new phe-nomenon with a scientific approach. The evolution of road traffic on an urban network remains to this day a poorly understood subject. The numerous intersections introduce strong correla-tions between the different roads, and the avail-able tools are quickly limited when one wishes to estimate the impact of a local decision on the network at a global scale.The field of traffic engineering is thus in a situation not unlike that encountered by the physics community when statistical physics was invented. Clearly, a ”microscopic” approach is not appropriate to the problem and a ”macro-scopic” point of view must be adopted. This new scale is typically that of the entire network of roads, ignoring the ”microscopic” reality of the behavior of motorists on these roads. In a physical approach, we try to find relevant aver-age quantities and describe their behavior. We focus in particular on three distinct aspects of the problem, each of which led to a publication. The first point is about the network itself, and in particular its evolution when the traffic de-mand increases. We show that traffic demand is the main parameter that led to the construction of urban freeways and ring roads in US cities during the 1960s. We empirically identify two population thresholds above which cities have a freeway and a ring road respectively and propose a cost-benefit analysis to understand this observation. The second point relates to the propagation of traffic jams on the urban net-work. We show that traffic jams occur simulta-neously on the entire network during peak hours. More precisely, we measure for each road its de-gree of saturation and ...