Modeling Vehicle Miles Traveled on Local Roads Using Classification Roadway Spatial Structure

This paper models the relationship between vehicle miles traveled (VMT) on local and collector roads with an objective to predict local road VMT by using collector road VMT. Through a continuous approximation method typically used for vehicle routing, it first analytically reveals this relationship mainly as a function of roadway density ratios between multiple roadway classifications. This structural relationship suggests regression equations using density ratios or logarithmic values of them as the explanatory variables. The use of regression equations enables to account for varying spatial distributions of roadways and demand through parameter calibration. The proposed regression equations are proved good fits through computer simulation using distinct community road network topologies. In addition, practical data from Hennepin County of Minnesota, U.S.A. that encompasses Minneapolis indicates that our developed regression equations can work well.