Development of a Generic Microscopic Cellular Automata Model on Traffic Flow

Abstract

Efficient design of traffic facilities catering to the safe movement of vehicular traffic can be achieved only by understanding the traffic stream properties in a comprehensive manner. In this thesis work, an attempt is made to closely observe traffic stream characteristics for various situations. Traffic stream properties is an outcome of human decision during driving. As human decisionmaking process is a complex phenomenon, understanding the stream characteristics using field experiments is a difficult task. To avoid this difficulty computer simulation of a model which is a powerful tool can be used to understand the traffic stream characteristics. In this thesis, an attemptis made to develop a microscopic simulation model using cellular automata (CA), which can simulate traffic stream of various densities, involving various road geometries, traffic composition involving homogeneous as well as heterogeneous traffic and movement characteristics of vehicles with and without lane discipline. The proposed model is a discrete microscopic model that describes movement of individual vehicle on a discrete representation of flow space and time. The flow space is divided into various square cells and time is divided into discrete time steps. Static and dynamic obstacles occupy flow space. Static obstacles refer to roadside vending, on street parking etc. whereas dynamic obstacles refer to other moving vehicles. Depending upon the type and position of the obstacle, the test vehicle takes decision about its course of action i.e., change in speed and change in steering angle at every time step. In order to complete the decision process every vehicle is assigned a pivot point (i.e., front right corner of every vehicle). It is with the reference to the pivot point every decision about the vehicle is taken i.e., calculation of speed, calculation of steering angle, updated position of the vehicle etc. A vehicle taking decision about changing its steering angle, first checks the relative speed (i.e. difference in speed between leading vehicle and following vehicle) and the space availability at its sides either to the right or to the left. The position of the pivot point of the following vehicle with respect to the position of the pivot point of the leading vehicle decides the steering angle change either to the left or to the right. Various microscopic and macroscopic studies were performed using the proposed model. Microscopic properties were studied for single lane traffic stream where the car following characteristics of the drivers was stressed. It may be noted that in case of single lane traffic stream study, the lateral movement of vehicles were absent. Microscopic studies include study of stability properties, time headway distribution and speed distribution. Macroscopic properties include relationship between speed flow-density for road widths starting from single lane to multi-lane and traffic composition with homogeneous and heterogeneous traffic. The proposed model takes into account the gap between the vehicles during motion as well as at static condition. Concept of buffer space has been proposed which takes care about the spacing between the vehicles. As in real life scenario, the gap between vehicles increase with increase in speed and is the least when at rest, the proposed model is capable of capturing the phenomena very well. Microscopic properties are also obtained from the proposed model which are an integral part of any microscopic traffic flow model. Collected field data (speed, flow and density) was used for calibration and validation of the proposed model. Parameters representing the gap between the vehicles during motion as well as at static condition (C1 and C2 respectively) are used for calibration. The values of C1 and C2 are changed iteratively until the square sum of errors (SSE) between the trendlines of distance headway - speed plots for simulated and field condition becomes the least. Statistical tests were performed to validate the model using the distance headway and speed data for simulated and observed data. Regression analysis for simulated distance headway versus field distance headway and simulated speed versus field speed has been done. The results obtained from regression analysis indicates the simulated data and observed data are matching well.