In this research, new methods to estimate vehicle miles traveled (VMT) for lower functional classes of roadways are introduced. The methods are based on the inherent correlation between VMT and roadway densities in each roadway class. This research found that the relationship between VMTs of different functional classes of roadways has to do with roadway typological structures according to functional classifications. To begin with, the analytical relationship between local VMT and collector road VMT was derived by assuming a grid network. The purpose was to find key relevant terms (basically roadway densities) in the relationship, which were used to define the format of regression equations. Next, the author proposed two types of regression models, one using density ratios as explanatory variables and the other using logarithmic value of roadway densities. Several simulation networks were set up to verify those proposed models using community road patterns categorized according to three different measures. The author found that the proposed models worked well for medium and high connectivity networks, but they were inadequate for simulating low connectivity networks. Moreover, the equation using logarithmic terms provided a better result in every numerical test. Next, the author verified the proposed regression equations in real situations. The results showed that the proposed regression models work very well in estimating urban local VMT of Minneapolis (grid networks). However, the relative error was much bigger in estimating local VMT of Bryan/College Station (non-grid networks). Finally, the author introduced a practical application procedure and also discussed the possible sources of errors in this study. This research introduces a potentially more efficient method (logarithm) for estimating VMT for lower functional classes of roadways.
