Modeling e-hailing and carpooling services in a coupled morning-evening commute framework
This project addresses several of Caltrans’ goals in its strategic plan: efficiency, sustainability, and accommodating and supporting innovative mobility technologies. The potential of integrating e-hailing, car-pooling and transit services seamlessly and more effectively could reduce solo-driving, and consequently lessen traffic demand, congestion, and vehicle miles traveled (VMT). A better understanding of this integration can also lead to better deployment of HOV lanes and car-pool pick-up and drop-off locations, to just name a few. The researchers will consult with the Office of High Occupancy Vehicle Systems (Matthew Friedman and David Liu) of the Division of Traffic Operations and Ms. Jeanie Ward-Waller of Caltrans’ Sustainability Program regarding the project from time to time.
The rapidly growing app-based transportation services, such as e-hailing services provided by Transportation Network Companies like Uber and Lyft, or casual car-pool enabled by SCOOP or WAZE apps, are transforming the travel behavior of individuals and urban mobility patterns. They also provide significant challenges to transportation planners and policy makers on how to assess the impact of these services on transportation systems, and how to facilitate or regulate these services because conventional planning tools are inadequate to model their more complex interactions between drivers, riders, and the private enterprises that link the drivers and riders together. In the conventional model, a trip is simply a vehicle with a driver traveling from an origin to an destination, but in an e-hailing service, additional trips can be generated to serve that one customer’s single trip, and because of this, such services could generate additional travel demand (in terms of number of trips) and VMT, as a recent study from New York city finds (it is estimated that e-hailing has increased the city-wide mileage about 14–19%. Clearly, one needs to have a better understanding of such services and develop new modeling tools that can capture the complex behavior of such services.
Regular car-pool (people paired up to drive together on a regular basis) or casual car-pool, on the other hand, does not have the serious deadheading issue (although even regular car-pool requires pick-up, the distance traveled in pick-ups is usually much shorter than the commuting distance). Its main problem, however, is with the often high “cost” for pairing car-poolers together. Here the term “cost” is used in a broad sense that includes money, time, inconvenience and other factors such as anxiety of riding with strangers. Because of this, the share of car-pool as a mode remains low despite the significant investment in car-pool infrastructure. In the commuting context, every car-pooler has a return trip, which makes the pairing even more challenging, and often hinders the use of car-pool. This, however, can be mitigated by integrating e-hailing and transit services with car-pooling. For example, a car-pooler for the morning commute can use an e-hailing service in the return trip, or use an e-hailing service to arrive at a transit station for the return trip if the e-hailing trip home turns out to be too expensive.
There have been extensive efforts to model these emergent services separately but many of these papers treat the ridesharing or pooling problem as a vehicle routing problem that does not take into account the congestion effects in transportation networks. The recent efforts to model several services together and take into account of the congestion effect are coming from this project team. For example, Ban et al 2019 integrates e-hailing service with solo-driving in a general network equilibrium context, and Ma and Zhang 2017 studied the integration of ridesharing in the morning commute from home to the central business district under dynamic ridesharing payment. There has been no effort, to the best of the researchers’ knowledge, to integrate both e- hailing and car-pool in a general network equilibrium modeling framework. In fact, a recent comprehensive review of various ridesharing literature reveals, among nearly a hundred and fifty publications, only one paper that attempts to account for the impact of ridesharing services in a general network equilibrium model, which was carried out by members of this research team.
This project will leverage the researchers’ understanding and previous work on e-hailing and ridesharing to develop a general equilibrium model that captures the complex interactions between solo-driving, e-hailing, car-pooling, and transit. In order to achieve this, the researchers will carefully examine the various factors that affect the demand and quality of service of e-hailing and car-pool, develop appropriate measures for them, and properly model them in a new framework that links both morning and evening commute together. The linkage is necessary because the choice of a travel mode in the morning, such as car-pool, can be affected by the availability of modes and choices made by other commuters in the evening return commute, which has rarely been studied in the literature. Last but not least, with the assistance of this new model, case studies will be carried out to understand how practices such as surge pricing in e-hailing services and factors such as mode-switching costs affect the overall performance of the network in terms of commuting cost, delay, and VMT. Such results can help Caltrans and other agencies such as MPOs make more educated decisions on promoting or regulating these emergent transportation services.
P.I. Name & Address
Funding source: Caltrans
Funding amount: $100,642.00
Start and end dates: 10/1/19 to 9/30/20