Calibration of Microscopic Traffic Simulation for Signalized Intersections Under Heterogeneous Traffic Conditions

One commonly used modeling approach for traffic systems is based on simulation. However, most of the traffic simulation solutions available are for homogeneous and lane-based traffic conditions. Some of them allow calibration of the model parameters to make it work for heterogeneous and lane-less conditions. One such solution, which is available as open-source, is SUMO (Simulation of Urban Mobility), a microscopic and continuous multi-modal traffic simulation package developed for homogeneous lane-based traffic. The present study focuses on calibrating SUMO for heterogeneous and lane-less traffic conditions. The study site selected is from Chennai, India. The calibration was done in two steps: in the first step significant parameters were identified and optimal values were found by trial and error. In the next stage, most influencing parameters were optimized using Genetic Algorithm (GA) by minimizing the error between the simulated and field travel time. Results showed an improvement of 82% compared to the default settings while simulating the heterogeneous and lane-less traffic conditions. The validated network can be used for various applications, including proper design and analysis of signals.