A New Arterial Free-flow Speed Model using Probe Vehicle Data

Free-flow speed is typically defined as the average vehicle speed under low-volume conditions with minimal influence on traffic control devices. It is arguably the most important input for arterial performance estimations and predictions. For example, free-flow speed is the baseline data for congestion estimates, air quality assessment, and travel forecasting. Consequently, the accurate estimation or prediction of free-flow speeds are crucial in the planning, operations, and performance evaluation of transportation systems.

The literature shows that the average free-flow speed on an arterial is a function of several geometric and roadway features including speed limit, access point density, median type, curb presence, segment length (i.e., signal spacing), bicycle lanes, etc. Because these factors vary over time and space, it is essential to develop a robust free-flow speed model that is based on real time conditions using historical data over different geographical locations. The recent editions of the Highway Capacity Manual (HCM) have in it a free-flow speed prediction model that is based on the posted speed limit and other roadway features. One key limitation of the HCM model is that it underestimates free-flow speeds for posted speed limits greater than 45 mph. The objective of this paper is to validate the HCM model using readily available probe vehicle speed data.

A preliminary probe speed data analysis on four arterials with varying posted speed limits (from 35 mph through to 50 mph) showed an improved model with a mean average percentage error of 0.4% compared to 6.3% when the HCM model was used. The detailed analysis of over 20 more corridors across the US is underway to develop a robust free-flow speed model under various roadway conditions.