Source-linked AI summary
On the inefficiency of ride-sourcing services towards urban congestion
Caio Vitor Beojone, Nikolas Geroliminis
TL;DR
Ride-sourcing can interfere with urban congestion, especially as fleet size and vehicle circulation change, but the relevant effects depend on how these services interact with other modes and passenger sharing. The paper combines taxi-trip data with an event-based simulation using an MFD traffic model to examine these effects and operational strategies. It finds that ridesplitting alone does not reduce vehicle-kilometers traveled, whereas restraining idle vehicles from cruising can reduce additional VKT.
Problem
The paper addresses limited understanding of how ride-sourcing operations affect congestion while replacing other modes and while idle vehicles cruise for passengers.
Method
The study combines taxi-trip data with an event-based simulation integrating a trip-based MFD traffic model, passenger–vehicle matching, and scenarios varying fleet size and willingness to share.
Results
Ridesplitting alone does not decrease VKT, congestion worsens as ride-sourcing fleets grow, and removing idle vehicles through parking strategies decreases additional VKT almost ten times.
Takeaways & Limitations
Reducing idle ride-sourcing vehicles is necessary to decrease VKT, while encouraging ridesplitting alone is insufficient.
Takeaways & Limitations
The analysis does not model an equilibrium game involving company incentives and pricing under congestion and empty-kilometer travel.
Abstract
from arXiv · showhide
The advent of shared-economy and smartphones made on-demand transportation services possible, which created additional opportunities, but also more complexity to urban mobility. Companies that offer these services are called Transportation Network Companies (TNCs) due to their internet-based nature. Although ride-sourcing is the most notorious service TNCs provide, little is known about to what degree its operations can interfere in traffic conditions, while replacing other transportation modes, or when a large number of idle vehicles is cruising for passengers. We experimentally analyze the efficiency of TNCs using taxi trip data from a Chinese megacity and a agent-based simulation with a trip-based MFD model for determining the speed. We investigate the effect of expanding fleet sizes for TNCs, passengers' inclination towards sharing rides, and strategies to alleviate urban congestion. We show that the lack of coordination of objectives between TNCs and society can create 37% longer travel times and significant congestion. Moreover, allowing shared rides is not capable of decreasing total distance traveled due to higher empty kilometers traveled. Elegant parking management strategies can prevent idle vehicles from cruising without assigned passengers and lower to 7% the impacts of the absence of coordination.
1 Introduction
Ride-sourcing has expanded urban mobility through app-based, on-demand transportation, while raising concerns about regulation, safety, pricing, and congestion. This paper investigates how fleet size, willingness to share, and operating strategies affect traffic conditions using a simulation framework that integrates matching with an aggregated traffic model.
- Transportation Network Companies use internet-connected mobile applications to match drivers and passengers in real time for on-demand transportation.
- Shared rides attempt to match passengers with similar trips within a time window, potentially increasing TNC and driver profits.
- TNC expansion raises concerns about unconstrained fleet sizes, absent price and service controls, privacy, safety, professional training, and surge pricing.
- Ride-sourcing may affect congestion differently depending on whether it replaces private vehicles, taxis, public transit, or induces latent demand.
- The paper tests fleet sizes, willingness to share, and operational strategies with an event-based simulation integrating a trip-based MFD traffic model and passenger–vehicle matching.
2 Data and Methodology
The study models ride-sourcing operations in Shenzhen by combining detailed taxi data with an event-based simulation and an MFD-based congestion model. It represents vehicles, passengers, matching, shared rides, traffic-dependent speeds, and parking strategies within the simulation.
- Data and traffic model: The Shenzhen dataset contains GPS coordinates for 199’819 trips and observations of 20’000 taxis every 30 seconds.
- Data and traffic model: The MFD computes network average speed from the accumulation of private vehicles and ride-sourcing vehicles under a homogeneous-congestion assumption.Shortest paths remain unchanged during simulation, reducing computational effort while representing trip-length distributions.
- Entities and vehicle states: The simulation includes private vehicles, waiting passengers, traveling passengers, and ride-sourcing vehicles with class-specific properties.Ride-sourcing vehicles track position, destination, activity, passenger count, and remaining distance, with a maximum capacity of two passengers.
- Entities and vehicle states: Ride-sourcing vehicles perform activities including cruising, parking, passenger pickup, single-passenger delivery, and shared-ride pickup and delivery.These activities describe vehicle state transitions during the simulation.
- Matching and ridesplitting: The matching process assigns nearby capable vehicles while enforcing passenger waiting-time, willingness-to-share, and detour requirements.Waiting passengers have 1 minute to receive an acceptable assignment before abandoning the request and choosing another mode.
- Parking strategy: The parking strategy assigns idle vehicles to lots near high-demand areas, where parked vehicles do not contribute to MFD accumulation.Parking lots are selected using a simplified two-stage p-median procedure.
3 Computational results
The simulations show that larger ride-sourcing fleets can worsen congestion, while parking idle vehicles improves service performance, traffic recovery, and accessibility. Ridesplitting alone does not reduce total vehicle travel, because empty-vehicle travel offsets shared-trip benefits.
- Overall findings: Ridesplitting does not decrease VKT, while growing ride-sourcing fleets worsen congestion.The results identify restraining idle vehicles from cruising as necessary to decrease VKT.
- Model validation: 2,000 real and simulated taxi-trip lengths were compared, and a 5% Kolmogorov-Smirnov test did not reject sample similarity.The authors conclude that the simulator represented trip lengths accurately despite shortest-path routing and regional demand aggregation.
- Service performance: Waiting times fell from 3.5–9 minutes with 1,000 vehicles to 0.4–1.5 minutes with larger fleets.Increasing fleet size improves service availability, but larger fleets also influence traffic speeds and journey duration.
- Parking strategy: Parking idle vehicles increased traveling speeds by 14% and shortened average journey duration by 1.4 minutes.The strategy prevented journey duration and waiting times from rising with larger fleets, but increased minimum waiting times by 1 minute on average.
- Traffic effects: With 3,000 vehicles, parking delayed reaching critical speed from 38 to 51 minutes and reduced hyper-congestion from 29 to 12 minutes.Parking idle ride-sourcing vehicles also accelerated post-peak recovery and improved system resilience.
- Traffic effects: Parking reduced additional VKT almost tenfold, while larger fleets increased VHT and reduced travel speeds.Negative additional VKT occurred only with small fleets, high willingness to share, and an active parking strategy.
4 Final considerations
The paper finds that ride-sourcing congestion depends on fleet operations and passenger sharing, with uncontrolled fleets and insufficient sharing increasing system impacts. Sustainable operation requires limiting empty cruising while encouraging ridesplitting.
- Sharing alone does not decrease system VKT when fleet size and operation remain uncontrolled.The paper links emissions reduction to changing TNC operations rather than relying on ridesplitting alone.
- TNC fleets should avoid cruising without assigned passengers to reduce vehicle kilometers traveled and emissions.
- Sharing reduces the number of vehicles needed to maximize coverage and minimize service times.
- When cruising cannot be avoided, greater willingness to share can minimize waiting and service times.
- A sustainable ride-sourcing service requires both removing passengerless vehicles from streets and increasing passenger receptiveness to ridesplitting.