Simulation-based joint optimization framework for congestion mitigation in multimodal urban network: a macroscopic approach

Abstract

Travel demand management (TDM) is an important measure that will aid in the realization of efficient and sustainable transportation systems. However, in cities where the most serious traffic congestion occurs, implementation of a single TDM measure might not be enough to reduce congestion, because the congestion mechanism in this case is highly complex and involves different transportation modes interacting with each other. Implementation of multiple TDM measures has rarely been discussed in the literature. Therefore, in this study, we propose a simulation-based joint optimization framework composed of dedicated bus lanes and vehicular congestion pricing. The objective of the optimization process is to minimize the congestion cost based on an advanced macroscopic flow theory called the multimodal macroscopic fundamental diagram (mMFD), which can capture the macroscopic traffic dynamics of multimodal transportation systems. In the proposed framework, we develop mMFD-based congestion pricing scheme and incorporate traveler’s behavioral model (i.e. joint departure time and mode choices) with the microscopic traffic simulator. We consider the Tokyo central area as a case study. The simulation results indicate that space allocation of 4.7% for the dedicated bus lanes would be optimal for Tokyo’s network, while the optimal congestion pricing scheme indicates that charges of 900 JPY between 7:30 and 8:00 AM and 300 JPY between 8.00 and 8:30 AM should be levied.

Appendices

Appendix A: Car and freight vehicle departure/arrival distribution

With regard to the general vehicles, the departure time distribution from the 23 special Tokyo wards (the special wards), the arrival time distributions for travel from the Saitama, Kanagawa and Chiba prefectures to the special wards and the arrival time distributions for travel from the other prefectures to the special wards are shown in Fig. 12a–c, respectively. In a similar manner, Fig. 12d–f show the corresponding distributions for freight vehicles. Obviously, these patterns are completely different. Therefore, by using these distributions for each demand, we can convert the daily coarse level demand into the hourly demand.

Fig. 12

a Departure time distribution from the special wards, b arrival time distribution for travel from the Saitama, Kanagawa and Chiba prefectures to the special wards, and c arrival time distribution for travel from the other prefectures to the special wards for general vehicles; d departure time distribution for travel from the special wards, e arrival time distribution for travel from the Saitama, Kanagawa and Chiba prefectures to the special wards, and f arrival time distribution for travel from the other prefectures to the special wards for freight vehicles

Appendix B: Groups of regions

The following table lists the groups of regions that were considered in this study. The group names correspond to those shown in Fig. 3 (Table 3).

Table 3 Groups of regions

Appendix C: Estimation of ASCs and the scale parameter for the departure time choice

To estimate the scale parameter and the ASCs for the departure time choice model, the following least squares method was used

$$ \mathop {\hbox{min} }\limits_{{\mu_{2} ,ASC_{{t_{i} }} }} J = \sum \left\{ {\left( {pax_{{b,t_{i} }} *D_{{b,t_{i} }} + 1.3*D_{{c,t_{i} }} } \right) - TD*\exp \mu_{2} U_{{t_{i} }} /\sum \exp \mu_{2} U_{{t_{j} }} } \right\}^{2} $$

(C-1)

where $pax_{{b,t_{i} }}$ is the average number of passengers per bus at time step $t_{i}$ for the result of the mode choice model, $D_{{m,t_{i} }}$ is the demand at mode $m$ at at time step $t_{i}$ from the simulation results, and $TD$ is the passenger total demand. Note that we assume that the vehicle occupancy of cars is 1.3 pax/veh in this study.