Operating speed prediction models for tangent sections of two-lane rural highways in Oklahoma State

In this study, geometric features (e.g. road and shoulder width), pavement characteristics (e.g. shoulder type, skid number, and international roughness index), and traffic parameters (e.g. posted speed limit and average daily traffic) were extensively analyzed to identify the significant predictors and develop better-fit 85th percentile speed prediction models for the tangent sections of two-lane rural highways. Three models, representing all highways, lower speed limit highways, and higher speed limit highways, were developed using stepwise multiple linear regression analysis. About 80% data from 251 locations in the State of Oklahoma, United States, was used in model development and 20% for validation. The all highways model is the best-fit among the three. Two most significant parameters of this model are posted speed limit and skid number. Additionally, international roughness index is significant in the higher speed limit highway model. The lower speed limit highway model depends only on the posted speed limit.