Estimation of Daily Traffic on Local Roads in Indiana State - A Mix of Heuristic Search Tool, GIS, and Visum Simulation

This paper presents a method of estimating the travel demand and vehicle miles traveled (VMT) in large regions including local roads and states. Verification of the procedure is developed for the Indiana Local Technical Assistantship Program (Tarko 2012). The model calibration and evaluation is presented for example parts of Indiana State. In the traditional approach, a network demand model includes properly calibrated traffic generation components applied to Traffic Analysis Zones and traffic exchange relationships between these zones calibrated based on household survey results. This paper attempts to calibrate a model that does not use the household survey but, instead, uses carefully chosen exploratory variables. The model is founded on a number of simplifying assumptions allowed by disaggregation of a large regional road network into much smaller parts called clusters. The network demand estimation could be then conducted for each cluster separately. The traffic generation parameters were fitted with a heuristic search method (genetic algorithm), and the values of exploratory variables (number of residents, employees, etc.) were obtained from microscopic GIS data. The calibrated model was evaluated by comparing the estimated daily traffic volumes with the values obtained from vehicle counting. This paper presents the model assumptions and the results obtained for a sample part of the road network.