Redefining ramp influence area for curved diverging and merging freeway segments using crash data.

Area of Science:

• Transportation Engineering
• Traffic Safety Analysis
• Roadway Geometry

Background:

• Freeway Ramp Influence Areas (RIAs) are typically defined by fixed distances, which may not accurately reflect heterogeneous crash patterns.
• Driver behavior, roadway geometry, and traffic operations significantly impact crash risk near ramps.

Purpose of the Study:

• To develop and validate a data-driven framework for empirically delineating RIAs based on crash data.
• To identify ramp-specific crash influence distances (τ) that account for geometric variations.

Main Methods:

• Utilized Negative Binomial Generalized Additive Models (NBGAMs) to model crash frequency based on ramp characteristics.
• Employed Negative Binomial gradient boosting (XGBoost) and change-point analysis with spatial bootstrap resampling to determine influence distances.

Main Results:

• Crash influence distances (τ) varied significantly by ramp geometry: curved entry ramps (≈1,800 ft), straight entry ramps (≈300-1,000 ft), curved exit ramps (≈500-900 ft), and straight exit ramps (≈300 ft).
• Demonstrated directional asymmetry: merging ramp disturbances diffuse downstream, while diverging ramp disturbances dissipate rapidly.
• Established a geometry-dependent hierarchy of ramp influence.

Conclusions:

• The proposed data-driven framework offers a more accurate method for defining RIAs compared to fixed buffer distances.
• Geometry-specific thresholds enable improved safety assessments and freeway ramp design.
• Supports flexible, risk-based RIA delineation for enhanced crash mitigation in complex environments.