Abstract

20 pages; International audience; This paper studies the pattern of departure times at a single bottleneck, under general heterogeneous preferred arrival times. It delivers three main outputs. Firstly, the existence of equilibrium is proven without the classical "S-shape" assumption on the distribution of preferred arrival time i.e. that demand, represented by the flow rate of preferred arrival times, may only exceed bottleneck capacity on one peak interval. Secondly, a generic algorithm is given to solve the departure time choice equilibrium problem. Lastly, the graphical approach that pervades the algorithm provides insight in the structure of the queued periods, especially so by characterizing the critical instants at which the entry flow switches from a loading rate (over capacity) to an unloading one (under capacity) and vice versa. Numerical illustration is given.

Abstract

This article models a novel driving-day-based tradable credit scheme (DD-TCS) to alleviate urban traffic congestion. In this model, car-using allowances (in terms of the number of days in a month, termed as “credit”) are freely and uniformly allocated to all travellers, who are also allowed to trade them in a market according to his/her travel needs (e.g. driving more or fewer days than the free endowment). As opposed to most studies on TCS, this paper explicitly considers the transaction cost (e.g. information cost of finding potential traders) in the trading market. To assess the feasibility of DD-TCS, we compare it against the license plate rationing (LPR) scheme, which has been practically implemented in many cities such as Beijing and Chengdu in China. Taking the performance of LPR as a benchmark, we quantify the threshold values of the transaction cost in DD-TCS when the two schemes yield equivalent performance (in terms of the total generalized cost). In numerical studies, we also compare the DD-TCS and LPR with the no-action case and the congestion pricing case (representing the theoretical optimum). Results show that both DD-TCS and LPR outperform the no-action case under certain conditions. With small transaction cost, DD-TCS may achieve a lower system cost that can be very close to the ideal optimum. In addition, parameter analysis shows that DD-TCS performs better than LPR in a wide range of transaction cost, where the threshold values appear to account for a considerable portion of the auto travel time. This implies that DD-TCS will be more appealing than LPR in practice because a transaction cost lower than the extremely large threshold values can be easily achieved for the trading market, e.g. via a mobile platform and modern communication techniques. Document type: Article