Promoting transit in the multimodal traffic environment

Abstract

Public transit is an effective method for combating traffic congestion in large cities where transit vehicles co-exist with other traffic modes such as cars and bicycles. However, present designs and operations of transit systems seldom consider the multimodal feature of urban traffic environment. In light of this, the present thesis explores optimal designs and efficient strategies for promoting transit systems. The study covers both the macroscopic (city-wide) and microscopic (city-block) scales. At the city-wide scale, we consider the design of an idealized transit network that can be accessed via shared bikes, since those bikes have become increasingly popular in many cities. Patrons may walk to shared-bike docking stations nearest their origins, and then cycle to their nearest transit stations where they deposit the bikes. The travel pattern is reversed when patrons cycle from their final transit stations on to their destinations. Patrons choose between this option and that of solely walking to or from transit stations. Shared bikes are priced to achieve the system-optimal assignment of the two feeder options. The idealized transit network is laid-out in hybrid fashion, where transit lines form square grids inside city centers, and radiate outward in the peripheries. A set of simplifying assumptions are adopted that pertain primarily to the nature of travel demand. These enable the formulation of a parsimonious, continuous model. The model produces designs that minimize total travel costs, and is ideally suited for high-level (i.e., strategic) planning. Similar models are developed for systems in which access or egress to or from transit can occur solely by walking, and by walking and riding fixed-route feeder buses in combination. Comparisons of these feeder options are drawn through numerical analyses. These are performed in parametric fashion by varying city size, travel demand, and economic conditions; and for trunk services that are provided either by ordinary buses, Bus Rapid Transit, or metro rail. Designs are produced for cases in which shared-bike and feeder-bus services are made to fit pre-existing and unchangeable trunk-line networks; and for cases in which trunk and feeder services are optimized jointly. Outcomes reveal that shared-bike feeder systems can virtually always reduce costs over walking alone, with cost savings as high as 7%, even when the shared bikes are made to fit a pre-existing transit network. Shared-biking often outperforms feeder-bus service as well. We further find that the joint optimization of trunk and shared-bike feeder services can reduce costs not only to users, but also to the transit agency that operates these services. Savings to the agency can be used to subsidize shared-bike services. We show that with or without this subsidy, shared-bike systems can always break even when they are suitably priced, and jointly optimized with trunk service. At the city-block scale, we consider a typical busy intersection approach controlled by a traffic signal, where buses and cars compete for green time and lane space. Signalized intersections are common bottlenecks in urban transportation systems where cars and buses may suffer from long queues and large delays. Existing bus priority strategies often promote transit at the cost of reducing cars' discharging capacity and creating more delays for cars. To solve this dilemma, we use a mid-block pre-signal that sorts through-moving and left-turning cars to increase their discharge capacity, and that also allows buses to skip the car queues. The pre-signal is further integrated with transit signal priority (TSP) schemes to further reduce bus delays. We develop a dynamic TSP strategy that determines the optimal TSP scheme to implement given the real-time bus arrival information. Analytical models are formulated to develop the car discharge capacity and the expected bus delay at the intersection approach. Trade-off between the above two metrics is examined. Numerical analysis is conducted for a wide range of operating settings with various bus arrival rates, numbers of lanes, and left-turning vehicle ratios. Our proposed intersection approach design with dynamic TSP strategy is compared against alternative designs. Results show that our design can greatly reduce bus delays without compromising cars' discharge capacity.