This paper studies the network capacity problem on signalized road network with reversible lanes. A Mixed Network Design Problem (MDNP) is formulated where the upper-level problem is a mixed integer non-linear program designed to maximize the network capacity by optimizing the input parameters, including the signal splits, circles, reassigned number of lanes and O-D demands, while the lower-level problem is the common deterministic user equilibrium assignment problem formulated to model the drivers' route choices. According to whether one way strategy is permitted in practice, two strategies for implementing reversible roadway are considered. In the first strategy, not all lanes are reversible and the reversible roadways always hold its ability to accommodate the two-way traffic flow. In the second strategy, one-way road is allowed. In this case, all the lanes in the reversible roadway could be assigned to one flow direction according to the unsymmetrical level of traffic flow in both directions. Genetic algorithm (GA) is detailed formulated to solve the bi-level network capacity problem. The application of the proposed method on a numerical example denotes that the network capacity with strategy 2 outperforms it is with strategy 1 considerably.
↧