Predicting interchange ramp volumes from interchange characteristics



Journal Title

Journal ISSN

Volume Title



Much of America’s freeway infrastructure is aging and will soon need repairs or rehabilitation. In the Dallas-Fort Worth region, in particular, a significant number of freeway interchanges soon need construction work. The City of Dallas is thus tasked with choosing the ideal times for ramp closures that would minimize congestion effects on the transportation network. Unfortunately, directly recording the traffic counts on these freeway interchange ramps is not trivial, so a method of imputing ramp traffic volumes is necessary. This thesis explores two main approaches of estimating ramp volumes based on count data provided by the Center for Transportation Research. The first approach relies on a multiple linear regression model that predicts ramp volumes based on main freeway lane volumes, interchange type, and variables to describe the traffic patterns exhibited on main lanes. The second approach computes a series of indices, or multipliers, for each hour of the day relative to an average hourly volume. To allow these indices to respond to various conditions, such as different days of the week or different types of interchanges, each of these multipliers is built into a multiple linear regression model. Throughout this research, a major hurdle is the lack of data recorded for the interchange ramps, so calibrating any type of model proves difficult. The second approach aims to reduce the errors caused by limited data by assuming that weekdays will exhibit very similar traffic patterns for most of the hours of the day. Due to its ease-of-use and overall performance, the second approach is recommended for use to estimate ideal times of day to close interchange ramps for repair.


LCSH Subject Headings