Modelling the Operating Speed in Segments of Two-Lane Highways from Probe Vehicle Data: A Stochastic Frontier Approach

Lobo, António; Amorim, Marco; Rodrigues, Carlos; Couto, António · 2018 · OpenAlex-citations

DOI: 10.1155/2018/3540785

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 lack of statistical models for predicting operating speeds in road segments, particularly on two-lane highways. While existing research predominantly focuses on "spot speed" models for individual design elements like curves and tangents, segment-level analysis is crucial for evaluating design consistency and defining speed limits during early planning stages. The authors introduce a novel model based on a Stochastic Frontier Approach (SFA) to predict any user-specified percentile speed in segments of two-lane highways, using probe vehicle data to account for driver behavior diversity and vehicle characteristics. The methodology utilizes data from nine road segments (totaling 23.5 km) on three National Roads in Portugal (N 14, N 101, and N 206). Speed data was derived from high-rate GPS probe vehicles, with 75 vehicles observed per segment in both directions, resulting in 675 total observations. Independent variables included geometric characteristics—bendiness, one-direction paved width (PW), extra lateral clearance (ELC), and density of intersections (DI)—as well as traffic volume represented by Annual Average Daily Traffic (AADT). The model builds upon the authors’ previous Operating Speed Frontier Model (OSFM), adapting it from spot speeds to segment speeds by aggregating geometric effects into a single scale factor and incorporating traffic and cross-section dispersion variables. The model was estimated using maximum likelihood methods, treating the maximum operating speed as a deterministic frontier and using an asymmetric disturbance term to account for speed reductions due to driver-vehicle diversity. The results indicate that geometric variables, particularly those related to cross-section width, have the most significant impact on operating speeds. The model successfully estimated elasticities for all variables, showing that increases in bendiness, intersection density, and traffic volume lead to speed reductions. For instance, a 10% increase in AADT under noncongested flow resulted in a 0.6% decrease in space mean speed. The model’s accuracy was validated by comparing its predictions against observed speeds and a traditional speed profile method. In seven out of nine calibration segments, the OSFM predictions differed from observed 85th-percentile speeds by less than 4 km/h, outperforming the speed profile approach. Further validation using simultaneous speed-traffic measurements on additional segments confirmed the model’s reliability. The significance of this work lies in providing practitioners with a macroscopic tool for evaluating design consistency and defining speed limits in the early stages of road planning, especially in rural areas where speed monitoring equipment is scarce. By allowing the estimation of any percentile speed, the model offers a more comprehensive understanding of speed distributions than traditional models that focus on single percentiles. This approach supports better infrastructure design and safety analysis by integrating both geometric and traffic factors into a unified predictive framework.

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.

• speed choice