Predicting 85th Percentile Speeds for Safer Roads: A Case Study Using Regression-Based Estimation

Abstract

Operating speed is one of the key parameters in road design, classification, and safety assessment. This study examines changes in 85th percentile (operating) speeds (V85) across the analyzed road network in the Middle East, between 2016 and 2020, using a combined dataset of measured and predicted values. The analysis covers 694 road segments, spanning approximately 1,949 km, across six functional road classes as defined in the relevant highway design manual. As a result of limited speed data coverage, a regression model was developed and calibrated using local speed survey data, geometric design attributes, and traffic volume classifications. This model enabled the generation of realistic, segment-level V85 predictions for use in road safety assessment tools such as the International Road Assessment Program. Results show a significant increase in operating speeds over the four-year period, particularly on higher-class roads such as rural freeways and urban expressways. Regression analysis revealed a strong correlation between V85 and design features including lane count, carriageway separation, and volume category, confirming international findings that drivers adjust behavior based more on roadway design than posted speed limits. Variability was also observed between urban and rural environments, as well as between divided and undivided roads. The study demonstrates the effectiveness of calibrated estimation models in contexts with incomplete speed data and highlights the importance of speed consistency assessment for safety planning. The findings support regional efforts to implement evidence-based speed management strategies and provide a replicable methodology for rapidly developing networks across the Middle East and similar regions.