Task Order 5402
Transit Operations Research

User Driven Scheduling of Transit Service

Samar Madanat
Civil and Environmental Engineering, California PATH
University of California, Berkeley

Summary

In this project, we seek to identify Demand Responsive Transit (DRT) service best suited for certain residents of Fremont and Newark, California.

The study can lead to improved transit service for these residents at lower cost. Further, it may prove itself an important step toward establishing general guidelines for a more systematic design of DRT. This is particularly important in light of the Americans with Disabilities Act.

Background and Methodology

We seek to design and field-test a DRT system to serve neighborhoods lying between Downtown Fremont and the City of Newark. The service would be used to make trips (including round trips) between private residences and popular destinations largely outside of these residential areas. Such destinations would include the Fremont BART station and AC Transit bus stops.

The work has two main components:

The first is a feasibility study. We will adapt mathematical models of DRT operation and then use them to evaluate service alternatives. The models will be used to assess feasibility of employing minibuses for DRT service and, pending a favorable outcome in this, will be used to identify optimal strategies. Models will also be used to determine the suitability of contract taxi service. From this, we might ultimately determine that minibus or taxi service should operate exclusively or perhaps instead, in concert together. One plausible strategy would be to use minibuses to serve "base" demands for service and then augment this with taxis when (and if) seasonal demand variations dictate need for greater capacities.

The second main component of the work is a field operational test. The form(s) of DRT identified as most suitable for the present locale (Fremont and Newark) will be deployed there, at least on a test basis. Performance (in terms of expected transit times for passengers) and costs (per passenger trip and/or per passenger mile) will actually be measured. Refinements to the DRT system may follow from this.

Work Plan

The first component of this work, the feasibility study, consists of three tasks (numbered 1 through 3 below). The second component, the field operational test, has two tasks (numbered 4 and 5). All tasks are briefly described next.

Task 1. Demand Estimation

Ridership counts collected by AC Transit, combined with standard statistical methods, will be used to estimate (time-varying) demands for DRT by origin and destination. These estimates will be essential inputs to the models adapted as part of the next task.

Task 2. Model Derivation and Evaluation of Alternatives

We will develop a mathematical model to describe DRT operation via minibuses. This model will estimate the amount of time passengers spend in the "system" (i.e., the sum of the time a passenger spends waiting to be picked-up and the time she actually spends in the minibus itself). Predictions will be based on factors such as demand for service, fleet size, routing strategy, etc. The model will thus determine the feasibility of DRT for the locale in question and, if feasible, the optimal service strategy (or strategies) as well.

The model will be of a continuum type; i.e., close approximations of passenger transit times will result from consideration of travel distances, largely independent of network geometry. To this end, we will modify proven continuum models for describing DRT service to many origins and to many destinations. We expect these existing models will require alterations because their assumptions are unsuitable for the proposed DRT's operating environment in two ways. First, most destinations served by the proposed system will likely lie outside the largely residential areas where trips originate. Second, the densities of these trip destinations will be lower than those of the origins.

Similar models will be used to evaluate contract taxi service. Berkeley researchers have been, of late, working on optimization problems for sizing fleets when minibuses and taxis operate jointly. This knowledge will be applied to the present feasibility study to evaluate the suitability of operating the DRT system with multiple modes.

Lastly, simple models of cost per passenger trip (and/or per passenger mile) will be developed. These models will also be used as part of our evaluations.

Task 3. Selection of DRT Mode(s)

With findings from the above tasks, we will furnish recommendations to Caltrans and AC Transit. Final decisions about the form of DRT to be field-tested will be made jointly with all partners in this project.

Task 4. Field Operations Test; set-up

In parallel with tasks 1, 2 and 3, we will perform the steps needed for the field operations test. These entail the development of the scheduling software (as outlined in the original proposal), acquisition of the communication systems (vehicle monitoring technologies, a computer-based reservation system for passengers, etc.). The system will be advertised within the community.

Task 5. Field Operations Test; measurement and evaluation

The DRT mode(s) selected in task 3 will be deployed and performance will be measured. These measurements will be extracted in automated (or at least semi-automated) fashion. They will include individual passengers' reservation times; their pick-up and delivery times; operating costs; etc.