Traffic congestion is increasingly becoming a severe problem in many large cities around the world. The problem is more complex in developing countries where cities are growing at a much faster rate than those in the developed world. Advanced Traffic Management Systems (ATMS) are one of the Intelligent Transport Systems (ITS) technologies that have been recommended and used as a tool to ease congestion problems in many large cities in the developing world. However, it is unknown how specific local conditions commonly observed in these cities, such as poor lane discipline and complex road user interactions, affect the performance of these systems. GETRAM (Generic Environment for Traffic Analysis and Modeling) was used in this research as a tool to develop microscopic traffic simulation models for the city of Bandung in Indonesia. The field data in this research, comprising throughputs, queue lengths and travel times, were collected during peak and off peak periods from all 90 signalised intersections connected to SCATS (Sydney Co-ordinated Adaptive Traffic System). This field data is believed to comprise one of the largest sets of “real world” data available for the development and validation of microscopic traffic simulation models. Two data sets were collected for this research: the first was used to develop and calibrate the simulation model and the second was used for validation. A number of statistical tests were used to determine the adequacy of the model in replicating traffic conditions. The results of statistical tests clearly showed that all of the calibrated and validated models reproduced field conditions with an acceptable degree of confidence. Therefore, the models were accepted as accurate and valid replications of the “real world”. The validated models were then used to evaluate the performance of SCATS which was implemented in Bandung in June 1997 as a pilot project. The results of comparative evaluation of the models under SCATS and under the Fixed Time control (without SCATS) demonstrated that SCATS did not necessarily always produce better results than the Fixed Time control. Furthermore, the performance of SCATS was strongly influenced by specific local conditions in the city. The multiple regression method was used to investigate the relationship between the traffic performance measures and significant basic variables. Based on this analysis, the main findings were: first, throughput was found to increase at intersections with higher v/c (volume to capacity) ratios. Second, throughput was found to decrease at intersections with higher numbers of phases and movements, longer widths of leg intersections, and farther distances to adjacent intersections. Third, queue length was found to increase at intersections with higher numbers of phases and movements. Based on the above findings, a number of improvements were recommended to enhance the performance of SCATS. This research also used traffic simulation to evaluate the impacts of these recommended improvements in increasing the performance of SCATS. The main findings from this evaluation were: first, restricted number of phases and movements at selected intersections substantially increased the traffic flow (78%) and decreased the queue length (by 55 to 67%) at the intersection. Second, making leg intersections wider—without physically building additional road capacity but by changing the stream with higher road hierarchy and higher v/c ratio from a two-way road into a one-way road—has a great impact on enhancing the performance of SCATS. Traffic flows were found to increase between 7 and 106%, and queue lengths were found to markedly decrease between 77 and 100% at all the suggested intersections. Third, the application of SCATS at intersections which are not closely spaced was not effective. Therefore, it is recommended that intersections which are not closely spaced remain under the Fixed Time control. The results and findings from this study provide road authorities in developing countries with an appreciation and enhanced understanding of the factors that influence the performance of traffic management systems in cities with similar characteristics to those in Bandung. These findings will also assist traffic engineers determine the best practices for the implementation of advanced traffic control systems in their cities.

New dynamic signal control methods such as fuzzy logic and artificial intelligence developed recently mainly focused on isolated intersection. In this study, a Fuzzy Logic Control for a Diamond Interchange incorporating with Fuzzy Ramp System (FLDI) has been developed. The signalization of two closely spaced intersections in a diamond interchange is a complicated problem that includes both increasing the diamond interchange capacity and reduce delays at the same time. The model comprises of three main modules. The Fuzzy Phase Timing module controls the current phase green time extension, the Phase Selection module select the next phase based on the pre-defined phase sequence or phase logics and the Fuzzy Ramp module determines the cycle time of the ramp meter bases on current traffic volumes and conditions of the interchanges and the motorways. The developed FLDI model has been compared with the traffic actuated simulation with respects to flow rates and the average delays of the vehicles. The model of an actual diamond interchange is described and simulated by using AIMSUN (Advanced Interactive Microscopic Simulator for Urban and Non-Urban Network) software. Simulation results show the FLDI model outperformed the traffic actuated models with lower system total travel time, average delay and improvements in downstream average speed and average delay.

Ramp metering, one of the most effective solutions for improving motorway traffic flows, is playing increasingly important role in traffic management systems. Because of its capability to handle nonlinear and non-stationary problems, fuzzy logic based ramp metering algorithms have been always considered as an extremely suitable control measures to handle a complex nonlinear traffic system. This thesis proposes a genetic fuzzy approach to design a traffic-responsive ramp control algorithm for an isolated onramp. For a local ramp meter algorithm, the problem could be described as the inflow optimization of on-ramp, based on the evaluation of motorway traffic condition. If the inflow of on-ramp is considered as the decision variable, the ramp control problem could be treated as a nonlinear optimization problem of maximizing the evaluation function. The adaptive genetic fuzzy approach is actually a control approach to maximize the inflow of on-ramp under the restriction of evaluation function. In this thesis, a well-known fuzzy logic based ramp metering algorithms developed by Bogenberger is introduced and implemented with an on-ramp congestion model of Constellation Drive Interchange in a stochastic microscopic traffic simulator, Aimsun. To improve the performance of fuzzy control system, genetic algorithm is applied to tune the parameterized membership function of each fuzzy input to maintain the flow density of motorway blow the estimated congestion density. The performances of the genetic fuzzy logic control ramp metering are compared with FLC (fuzzy logic control) ramp metering by means of the percentage change of TTT (Total Travel Time) based on no control condition in Aimsun. The simulation results show the genetic fuzzy ramp metering has a more significant improvement on TTT and more strong stability to maintain system flow density than FLC ramp metering.