Improving Simulation Models of Unsignalized Intersections on Rural Multilane Highways through Calibration Using Naturalistic Driving Study Data from Strategic Highway Research Program 2

Abstract

The aim of this study is to create a calibration method for traffic simulation models of two crucial movements at unsignalized intersections on rural divided highways. This analysis draws on data from the Naturalistic Driving Study, specifically direct left turn (DLT) movements at conventional intersections and right turn followed by U-turn (RTUT) movements at restricted crossing U-turn (RCUT) intersections. Using 453 naturalistic trips from 79 drivers, the speed trajectory, stopping behavior, lane utilization, and acceleration and deceleration rates were analyzed. Three-step and four-step calibration methods were then developed for the Vissim simulation models for DLT and RTUT movements, respectively, which were verified by comparison with naturalistic driving data. The calibrated Vissim models were verified by comparing the naturalistic driving speed trajectories of DLT and RTUT movements. The results show that calibrated models can improve the operational analysis for overall unsignalized intersections on rural multilane divided highways.