EMPIRICAL ANALYSIS AND MODELING OF FREE-FLOW SPEED FOR MULTILANE HIGHWAYS Piyamas Suthida Chansirikul

This study developed and evaluated free flow speed (FFS) estimation models for multilane highways in Thailand by applying multivariable linear regression. Data from 272 highway segments were analyzed considering the percentage of heavy vehicles, left shoulder width, and access point density. Highways were categorized into four-lane and more-than-four-lanes configurations and further analyzed under different base free flow speeds (BFFS) ranging from 90 to 120 km/h. For four-lane highways, Model 1-2 (BFFS = 100 km/h) achieved the best predictive performance, with an R² of 0.9583, root mean squared error (RMSE) of 1.51, and mean absolute error (MAE) of 1.24. For more-than-four-lanes highways, Model 2-3 (BFFS = 110 km/h) performed best (R² = 0.9387, RMSE = 1.86, MAE = 1.46). All variables showed significant negative effects on FFS. The developed models provide practical tools for assessing roadway performance, thereby regulating speeds, and guiding infrastructure planning, enhancing safety and efficiency on Thai multilane highways