Comparison of Speed-Density Models in the Age of Connected and Automated Vehicles

Karbasi, Amir Hossein; Mehrabani, Behzad Bamdad; Cools, Mario; Sgambi, Luca; Saffarzadeh, Mahmoud · 2023 · Crossref

DOI: 10.1177/03611981221118531

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the need for accurate speed-density models capable of accounting for Connected and Autonomous Vehicles (CAVs) in traffic flow analysis. While traditional models like Papageorgiou’s do not incorporate CAV penetration rates, recent models such as Lu et al.’s semiparametric approach do, but have only been validated in urban networks with aggressive driving behaviors. This paper compares the estimation power of the Lu model against the established Papageorgiou model in both freeway and grid networks, evaluating three distinct CAV driving behaviors: aggressive, normal, and conservative. The researchers employed microscopic traffic simulation using the open-source SUMO software to generate Fundamental Diagrams (FDs). They utilized the Krauss car-following model and the LC2013 lane-changing model, modifying parameters to reflect HDV and CAV behaviors based on literature from Atkins Ltd and other studies. Specifically, CAV behaviors were defined by varying time headways, minimum gaps, acceleration, and lane-changing assertiveness. Simulations were conducted on a 5 km two-lane freeway (100 km/h speed limit) and a 49-intersection grid network (50 km/h speed limit). CAV penetration rates were incremented in 20% steps from 0% to 100%. The resulting simulation data were fitted to both the Lu and Papageorgiou speed-density models. Model performance was evaluated using Mean Absolute Percentage Error (MAPE) and t-tests, while capacity impacts were derived from the generated FDs. Results indicate that both models accurately estimate speed-density relationships, with R² values ranging from 0.89 to 0.92 for the Lu model and 0.89 to 0.91 for the Papageorgiou model. Statistical tests revealed no significant differences between the capacities estimated by the two models or between model estimates and raw simulation data. However, CAV driving behavior significantly influenced road capacity. In freeway scenarios, aggressive CAV behavior increased capacity by up to 101%, normal behavior increased it by up to 30%, and conservative behavior decreased it by up to 34%. The Lu model’s coefficients confirmed that aggressive and normal CAV behaviors positively impact average speed and mitigate the negative effects of density, whereas conservative behaviors have the opposite effect. Although the Papageorgiou model showed slightly lower MAPE for speed estimation, the differences were negligible. The study concludes that the Lu model is a robust tool for estimating speed in mixed traffic environments, performing comparably to the well-established Papageorgiou model even in high-speed freeway networks. Crucially, the findings demonstrate that the impact of CAVs on traffic capacity is highly dependent on their driving behavior; aggressive automation yields substantial capacity gains, while conservative automation can reduce capacity below baseline levels. This highlights the importance of defining specific CAV behavioral parameters when assessing the operational benefits of autonomous vehicle integration.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density