Horizontal Curve Safety Performance Evaluation Based on the Naturalistic Driving Study Lane Position Data

Claros, Boris; Chitturi, Madhav V.; Vorhes, Glenn; Bill, Andrea; Noyce, David A. · 2024 · ROSA P / United States. Federal Highway Administration. Office of Safety and Operations Research and Development

archive: archived pipeline: cataloged verified

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

Summary

This research addresses the lack of resources for transportation practitioners to evaluate roadway design based on safety performance, particularly for rural, undivided, two-lane horizontal curves. Conventional safety analysis relying solely on observed crash data is often insufficient due to limited data availability and the inability to quantify the specific effects of performance-based design (PBD) implementations. To overcome these limitations, the study utilizes safety surrogates derived from naturalistic driving data to estimate the safety impact of geometric elements that influence driver behavior. The primary goal was to develop an analytical tool that allows practitioners to perform economic assessments of curve design alternatives by quantifying tradeoffs between safety improvements and implementation costs. The methodology combined data from the Roadway Information Database (RID) and the Second Strategic Highway Research Program Naturalistic Driving Study (SHRP2 NDS). Researchers initially identified 43,296 curves in the RID, filtering them down to 3,292 rural, undivided, two-lane curves with grades between -1% and +1% and radii between 250 and 3,000 feet. These curves had 150,233 associated vehicle traversals. A rigorous data processing protocol was applied to ensure quality, resulting in a final dataset of 2,030 curves and 62,414 traversals. This process involved removing curves with intersections, divided roadways, rumble strips, or inconsistent geometry, as well as traversals with lead vehicles, missing data, or low-probability lane position measurements. Lane position data proved to be the most consistent and statistically significant safety surrogate. Researchers used negative binomial regression modeling to predict centerline and edge-line encroachment frequencies based on curve geometry, such as radius and shoulder width. These encroachment estimates were then associated with observed crash data from state records to convert surrogate events into predicted crash rates. The results demonstrated that predictor variables like curve radius and shoulder width significantly influenced safety outcomes. Specifically, predicted crashes decreased as curve radius increased, and similarly decreased as shoulder or lane width increased. The study established a reliable association between lane departure encroachments and observed crashes, validating the use of these surrogates for safety estimation. Using these findings, the researchers developed an analytical tool for practitioners. This tool accepts inputs such as curve radius, shoulder width, lane width, length, annual average daily traffic, construction cost, and expected service life. It calculates estimated crashes and associated costs, enabling a benefit-cost analysis of different design configurations. The significance of this work lies in providing a quantitative, performance-based approach to curve design that moves beyond traditional crash data limitations. By linking naturalistic driving behavior to crash outcomes, the study offers a robust method for evaluating the safety efficacy of geometric design elements. The resulting analytical tool empowers roadway designers and safety professionals to make informed decisions by explicitly assessing the economic tradeoffs between safety benefits and implementation costs. This supports the broader adoption of performance-based design practices, ensuring that infrastructure projects meet specific safety goals while optimizing resource allocation.

Key finding

Lane position encroachments served as the most consistent safety surrogate, with negative binomial models showing that predicted crash rates decrease as curve radius and shoulder or lane widths increase.

Methodology

naturalistic

Sample size: 150233

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. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

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

• road geometry
• incidence prevalence
• naturalistic crash near crash
• lane positioning
• perceptual countermeasures

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: crash risk outcomes, observational prevalence
• Methodological Resource: dataset resource