[Paper Review] Identifying the threshold to sustainable ridepooling

Charlotte Lotze, Philip Marszal+4 more|arXiv (Cornell University)|Jun 9, 2023

Transportation and Mobility Innovations4 citations

TL;DR

This paper introduces a dimensionless load parameter that quantifies the sustainability threshold of ridepooling services by balancing demand and supply timescales. It predicts the break-even point where ridepooling drives less total distance than individual car trips, enabling a priori assessment without simulations or operational data.

ABSTRACT

Ridepooling combines trips of multiple passengers in the same vehicle and may thereby provide a more sustainable option than transport by private cars. The efficiency and sustainability of ridepooling is typically quantified by key performance indicators such as the average vehicle occupancy or the total distance driven by all ridepooling vehicles relative to individual transport. However, even if the average occupancy is high and rides are shared, ridepooling services may increase the total distance driven due to additional detours and deadheading. Moreover, these key performance indicators are difficult to predict without large-scale simulations or actual ridepooling operation. Here, we propose a dimensionless parameter to estimate the sustainability of ridepooling by quantifying the load on a ridepooling service, relating characteristic timescales of demand and supply. The load bounds the relative distance driven and uniquely marks the break-even point above which the total distance driven by all vehicles of a ridepooling service falls below that of motorized individual transport. Detailed event-based simulations and a comparison with empirical observations from a ridepooling pilot project in a rural area of Germany validate the theoretical prediction. Importantly, the load follows directly from a small set of aggregate parameters of the service setting and is thus predictable a priori. The load may thus complement standard key performance indicators and simplify planning, operation and evaluation of ridepooling services.

Motivation & Objective

• Address the lack of predictive tools to assess whether ridepooling services will be more sustainable than private car use.
• Overcome the limitations of standard key performance indicators like average occupancy or relative distance driven, which are hard to predict a priori.
• Develop a theoretically grounded, parameter-based metric that estimates the break-even point for ridepooling sustainability using only aggregate system parameters.
• Validate the model across simulations and real-world data from a rural German ridepooling pilot project.
• Provide a scalable, system-level indicator to support planning, evaluation, and comparison of ridepooling services across different urban and rural contexts.

Proposed method

• Define a dimensionless load parameter as the ratio of the average request rate to the average vehicle velocity, scaled by the average direct trip distance.
• Derive a theoretical inequality linking the load to the relative distance driven, showing that sustainability improves when the load exceeds a critical threshold.
• Use event-based simulations to test the load’s predictive power across diverse demand and supply conditions.
• Validate predictions against empirical data from a real-world ridepooling pilot in rural Germany, comparing simulated outcomes with actual service performance.
• Extend the model to include realistic factors such as stop times, vehicle capacity limits, and non-uniform request distributions.
• Demonstrate that the load captures long-term system dynamics and serves as a robust indicator of sustainability potential under steady-state conditions.

Experimental results

Research questions

• RQ1 At what level of service demand does ridepooling become more sustainable than individual car use in terms of total distance driven?
• RQ2 Can a simple, aggregate, and predictive metric be derived to estimate ridepooling sustainability without relying on large-scale simulations or operational data?
• RQ3 How does the load parameter relate to standard key performance indicators like average vehicle occupancy or relative distance driven?
• RQ4 To what extent does the load parameter remain predictive under varying demand patterns, vehicle speeds, and service configurations?
• RQ5 Can the theoretical load threshold be validated using real-world ridepooling data from a rural mobility pilot project?

Key findings

• The proposed load parameter uniquely identifies the break-even point where the total distance driven by a ridepooling fleet equals that of direct individual car trips.
• The load is a dimensionless, system-level metric derived from only three aggregate parameters: request rate, average direct trip distance, and average vehicle velocity.
• Theoretical predictions based on the load closely match results from detailed event-based simulations across various demand and supply scenarios.
• Empirical data from a rural German ridepooling pilot project confirms the model’s predictive accuracy, even under non-ideal and variable conditions.
• The load remains robust even when accounting for realistic complexities such as stop times, limited vehicle capacity, and non-uniform request distributions.
• The load enables a priori evaluation of ridepooling sustainability, offering a practical tool for planners and operators to compare services across different settings without simulation or operational history.