Parking Search in Urban Street Networks: Taming Down the Complexity of the Search-Time Problem via a Coarse-Graining Approach

Abstract

The parking issue is central in transport policies and drivers' concerns, but the determinants of the parking search time remain relatively poorly understood. The question is often handled in a fairly ad hoc way, or by resorting to crude approximations. Very recently, we proposed a more general agent-based approach, which notably takes due account of the role of the street network and the unequal attractiveness of parking spaces, and showed that it can be solved analytically by leveraging the machinery of Statistical Physics and Graph Theory, in the steady-state mean-field regime. Although the analytical formula is computationally more efficient than direct agent-based simulations, it involves cumbersome matrices, with linear size proportional to the number of parking spaces. Here, we extend the theoretical approach and demonstrate that it can be further simplified, by coarse-graining the parking spot occupancy at the street level. This results in even more efficient analytical formulae for the parking search time, which could be used efficiently by transport engineers.