The main objective of the thesis was to develop and apply methodologies for analysis and evaluation of the applicability of different types of road traffic simulation models for urban traffic conditions. For this purpose a number of models of macroscopic, mesoscopic and microscopic type were applied, calibrated, validated and analyzed using extensive field data collection to obtain traffic flow, travel time and queue-length data for major road networks in Stockholm and Uppsala. The thesis is composed of four parts. Part I contains the background statement, objectives of the study, overview of modeling techniques, synthesis and recommendations for further research. Part II to Part IV document separate studies with different traffic models including objectives, model description, data collection, analysis and conclusions. The titles convey the specific type of traffic facility selected and investigated using various transport modeling techniques. Part II: Urban Arterial Study treats the subject of calibration and validation of the stochastic microscopic traffic simulation model TRAF-NETSIM for Swedish road traffic conditions. The method is described and the sensitivity of the distributions of free flow speed, start-up lost time, queue discharge headway and accepted left turning gaps is described. It was shown that the quality of the model improved and that it is applicable for local traffic conditions. Part III: Urban Traffic Corridor Study describes the implementation; calibration and analysis performed on a number of high-resolution mesoscopic and microscopic (CONTRAM and CORSIM) transport models, a prerequisite for the need to model fast emerging ITS systems. Field surveys feed the models and expose their ability to reproduce the actual conditions observed in the field. Both models reproduced the field data satisfactory. The extent of the network imposed limitations on the operation of the microscopic model. Part IV: Urban Road Network Study investigates the hypothesis that the use of macroscopic transport planning models may cause significant errors if applied on congested networks. An alternative method to analyze congestion impacts is the use of dynamic meso- or microscopic simulation models. Two macroscopic static assignment models, DSD and TransCAD, and the dynamic assignment model CONTRAM were tested and comparative studies performed based on comprehensive field data. The macroscopic models were shown to be poor indicators of network travel time. Better correlation was obtained from the mesoscopic model. (Author/publisher).
Abstract