Evaluation of Different Vehicle Following Models Under Mixed Traffic Conditions

AbstractCar-following models replicate the behavior of a driver following another vehicle. These models are widely used in the development of traffic simulation models. Only fewer studies have been conducted to compare the performance of different car following models under mixed traffic conditions. The present study focuses on the evaluation of different car following models under mixed traffic conditions. Specifically, the following four cars following models were selected: 1. Gipps Model, 2. Intelligent Driver Model (IDM), 3. Krauss Model and 4. Das and Asundi Model. These models were implemented in a microscopic traffic simulation model for a mid block section. Each of these models is then calibrated for three states: non-steady state with constant parameters across classes, steady state parameter and non-steady state with classwise parameters. Then the models are evaluated using the performance measures such as error in hourly stream speeds and classwise speeds, critical parameters and Mean Absolute Percentage Error (MAPE) for speed and density values, obtained at one minute intervals.The main contribution of this paper is calibration of models using different Measure of Effectiveness (MoE) simultaneously. This better replicates field observed conditions compared to calibration based on single MoE. From the analysis, it is observed that non-steady state performs better than steady state. The performance of non- steady state class wise is better compared to non-steady state with constant parameters based on the performance measures. Among the four models, Gipps model is able to replicate the field conditions better than other models in non-steady state. The findings have interesting implications for capacity and PCU estimation and Level of Service (LoS) Analysis