Inhomogeneous flow patterns in undersaturated road networks

Travel conditions on a road network are described by a macroscopic fundamental diagram (MFD) only when traffic is homogeneously distributed over all of the network's links. Otherwise, a network's measured traffic states fall beneath its MFD, and network flow therefore falls below the maximum. The present study found that inhomogeneous link flowed in a network may emerge before the beginning of a peak period, even before persistent queues had formed. This finding is potentially problematic because it is known that inhomogeneities of this kind can persist in a network for long durations. Therefore, the inhomogeneities that are predicted for undersaturated traffic conditions may persist as the network moves to the brink of congestion, and even well beyond. If drivers are left to their own devices, a network will never serve its peak period demand at the maximum rate. Steady state conditions in undersaturated networks of signalized intersections were modeled in simple ways. Analyses were performed for an idealized street grid characterized by uniform origin-destination patterns and a more realistic demand pattern that resembles one in downtown San Francisco, California. Emergent patterns in inhomogeneous link flow were examined. Preliminary ideas to combat these undesirable patterns are discussed.