Abstract:

Population of Karachi is growing at a phenomenal rate. This increasing population become one of the factors of increasing traffic. Proper traffic management through various tools such as formulation of models and their testing in different scenarios for evaluation of strategies, implementation of Intelligent Transport System based solutions are vital to improve the current situation. Almost all Traffic microsimulation packages and models are developed for conditions prevailing in developed countries; their direct applications in testing different intervention for developing countries may be questionable. Therefore, as a first step it is important to adapt these models (such as their calibration) for the traffic behaviour prevailing in developing countries. Based on this, the aim of this research is to calibrate the car-following models for traffic stream scenario of Karachi city.

There are many car-following models presented in the literature but this research focused on only two car-following models i.e. General Motor's fifth model and Modified Pitt Car-following model. This thesis reported calibration of the car-following models parameters such as sensitivity (K), Reaction time (R), speed exponent (m), distance headway exponent (ɭ), , and time spacing (Ts). For calibration, field data trajectories of vehicles limed on three parameters such as position, velocity and acceleration has been obtained from the collected traffic video data of two locations i.e. Disco Bakery and M.A Jinnah Road. Based on this, three different scenarios are examined which are at intersection for Signal operating and Signal non-operating condition and at midlink. The extracted data is en used as an input for models and then Root mean square error (RMSE) technique is med to calibrate the models parameters. The minimum values of RMSE have been achieved for all the three scenarios using a heuristic method which is programmed in MATLAB. A range of RMSE from O to 20 has been assumed acceptable and values of the calibrated parameters which are achieved at defined range of RMSE have been recommended for three different scenarios. Furthermore, the recommended calibrated parameters values are also supported through statistical analysis (such as t-test and F-t¬est). Additionally, acceleration trajectories obtained from field data and modelled data is compared and based on that it has been found that General motor's model perform well for Scenario 1 and 3 and for Scenario 2 Modified Pitt-car following model behaves appropriately