Simulation Model Finds That Pooled First-Mile, Last-Mile Transit Access Services Reduce VMT and Travel Time and Increase Revenues Compared to Single-Passenger Service.

San Francisco Bay Area model simulates an AV fleet that offers first-mile access to transit at different price points for three different service models.

Date Posted

06/28/2021

San Francisco, California, United States

San Francisco Bay Area: San Francisco, California,

United States

Identifier

2021-B01576

Cost-Benefit Analysis of Novel Access Modes: A Case Study in the San Francisco Bay Area

Summary Information

This study set out to model and simulate the travel and revenue impacts of a fleet of automated vehicles (AVs) as a potential solution for the first-mile problem of transit access to help lower barriers to potential riders in the San Francisco Bay Area.

The simulation was studied over three different scenarios, each offering a range of fares:

Pick Up and Drop Off Points (PUDO), which involves eight-seat automated vehicles providing a ridesharing service that picks up riders who walk from the transit drop off point to their home.
Door-to-door, single-passenger ride hailing service (D2D Single).
Door-to-door ridesharing (D2D Ridesharing), where eight-seat automated vehicles provide a home-based pickup ridesharing service.

A multi-agent-based simulation model was used and the effects of AVs on vehicle miles traveled (VMT), travel time, operation cost, and revenue were estimated.

Methodology

The study used BA-MATSim, a Bay Area simulation model with a module that simulates the effects of automated taxi fleets. The algorithm dispatches a fleet of empty, driverless vehicles to fulfill pickup requests, with vehicles selected based on proximity. The research team created a fleet size of 5,000 vehicles that started their day randomly throughout the model area. Pickup and drop-off locations were determined by an optimization framework in which walking and wait times are minimized.

In combination with AV fleet dispatch optimization, different pricing was applied to the automated taxis to determine the impact of fare differences on ridership and revenue (pricing ranged from $0 to $10 per ride based on current prices for microtransit services). Revenue derived from the automated taxis were compared to the costs of operating the automated taxis per mile, using an estimated range of costs from $0.30 to $0.50 per mile.

Findings

The study found that $2–$3 per ride is the optimal price for revenue generation based on the time and monetary costs of the transit riders in the study area.
AVs employed as a transit access service are faster (i.e., have a lower travel time for users) and have higher ridership when the rides are less expensive (less than $3 per ride).
Pooled AV access services can reduce VMT by up to 50 percent and can reduce zero passenger miles (i.e., number of miles traveled by the AVs with no passengers) by up to 65 percent when compared with single-passenger AV ride hailing service when the fare is low.
D2D-Rideshare services can generate up to $200,000 in revenue per day when the fare is $3 per ride. When the operational cost is considered, pooled service models are found to be more likely to create profits, while single rider models may not. D2D-Rideshare was found to yield the highest profit regardless of fare cost.