Calibration of VISSIM Parameters for Modeling Heterogeneous Traffic Conditions in Kathmandu using Genetic Algorithm: Case Studies of Singhadurbar, Padmodaya, and Putalisadak Intersections

Abstract

Most developing countries consists of unsignalized intersections carrying heterogeneous traffic conditions. Microsimulation models and their environment are widely being used for the evaluation of such traffic conditions and development of road networks worldwide. Due to the difficulties in analyzing the complexities of heterogeneous traffic and calibrating the microsimulation, their use in Nepalese context is limited to research studies. This study proposes a procedure to determine the range of values of sensitive calibration parameters tailored to local conditions. Additionally, it suggests a methodology for the automated calibration of microscopic simulation models based on these sensitive parameters at unsignalized intersections with heterogeneous traffic. VISSIM, a microscopic, time step oriented, and behavior-based simulation tool was used to model the intersection which consists of a large number of input parameters making model calibration rather difficult. Sensitivity analysis was performed to identify the sensitive calibration parameters using Latin Hypercube Sampling (LHS) and one way ANOVA testing. Based on the findings of the literature review, 12 calibration parameters were identified. Using traffic flow as a measure of effectiveness, the parameters were reduced to 9 sensitive calibration parameters.