Thesis (M.Eng.) - Asian Institute of Technology, 1998

A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Civil Engineering

Chaos theory has been proved very useful in exploring the behavior of a number of complex dynamical systems such as weathers forecasting, fluid dynamics and many other complex systems. It is impossible to make long-term predictions about the systems with chaotic behavior albeit this theory can provide guidance for short-term predictions under certain circumstances and also can contribute in providing insight into the complex behavior of the system. Traffic flow is a complex phenomenon because of its nature to undergo sudden changes especially when the traffic condition is congested due to irregular stop-start situation and many other contributing factors such as drivers behavior, driving enviromnent, road and traffic conditions etc. As a vehicle must have unique response for a particular traffic condition, the traffic flow can be treated as a deterministic process so it is possible that the complexity in this system may have arisen from the presence of chaotic behavior in the motion of vehicles. The principal goal of this study is to apply chaos theory in traffic flow to analyze the motion of vehicles in traffic stream. In this study a microscopic simulation program was developed using collision avoidance car following model that animates the movements of each vehicle in the platoon. This program was extended to produce time series diagrams, phase diagrams, Poincare sections, power spectrum which were used to analyze the sensitivity of each of model parameters and as indicators of presence of chaotic behavior in the system. The values of Lyapunov exponent for each vehicle was calculated to characterize their motion and this along with the value of KS entropy were used to measure the predictability of traffic flow conditions. It was found that for some particular initial conditions, some of the following vehicles in the platoon have chaotic motion when a periodic disturbance was introduced to the equilibrium state of car following model but still the motion of vehicles can be predicted for short duration of time.