Lead Presenter Biography

David is in his fourth year of transportation demand modeling and brings valuable outside perspective and technical skill to the field. He has a demonstrated ability for complex problem solving as shown by his doctoral work in nuclear physics and is a well-versed programmer, particularly in Python and C++. His primary focus is on travel demand forecasting and has contributed to the development of travel demand models in many regions including San Francisco Bay, Detroit, and San Diego. He is also an active software developer on the ActivitySim activity-based modeling platform.

Co-Authors

Presentation Description

Transit ridership is an essential part of any travel forecasting model. Model parameters determine how transit ridership responds to changes in policy, land use, and network, and generally include quantities like travel time, travel cost, wait time, and access/egress time. Often overlooked aspects of transit include the user experience with respect to the number of people on the transit system, the capacity of the transit system, and the reliability of transit operations. For regions with very crowded transit systems, these effects can be significant.

To better model transit in the nine-county San Fransisco Bay Area, transit capacity, crowding, and reliability has been included in the activity-based Travel Model 2 (TM2) forecasting model. TM2 is maintained by the Metropolitan Transportation Commission (MTC) (the transportation planning, financing, and coordinating agency for the nine-county San Francisco Bay Area) and is built upon the Coordinated Travel – Regional Activity-based Modeling Platform (CT-RAMP) platform. TM2 uses the commercial Emme software to perform capacitated transit assignment and capture the transit system markers in network skims which are then fed back to the mode choice model.

In the enhanced model, individual transit line capacities used in assignment restricts maximum number of riders for each transit line. High transit volume to capacity ratios also increases the effective headways of transit lines due to people having to wait more the next available vehicle with space. Transit crowding is calculated based on the volume to capacity ratio and considers the number of seated and standing spaces. A penalty is applied in assignment to make crowded paths less attractive and is perceived as a factor on the in-vehicle transit times. Transit capacity is included by restricting the availability of parking at transit stops according to the acquired parking lot data in the region. A novel simulation-based approach was developed in which drive-transit tours arriving at over-capacity transit station parking lots are re-simulated after removal of the full lots.

Transit reliability is represented by two separate calculations. First, reliability related to the transit stops is calculated based on the number of people boarding the transit line. More boardings can lead to greater disruption and a decrease in reliability of the service. Second is the reliability of the transit vehicles to get from one stop to another. The more urban the route and the more volume on the road, the greater the likelihood for delay, and therefore a decrease in reliability.

Transit crowding, capacity, and reliability penalties are skimmed from the transit network and fed back into the mode choice model to allow for switching between transit different paths or away from transit all together.

The presentation will discuss the implementation of transit capacity, crowding, and reliability in the San Fransisco Bay area’s activity-based model, show the transit assignment results, and discuss the sensitivities of the approach for use in policy decisions. 

Presentation File