Multiclass multilane traffic models and integrated optimal control strategies for freeways

Abstract

Active Traffic Management (ATM) relies on a fast and reliable traffic simulator, for the rapid quantitative assessment of various control strategies under different traffic scenarios, such as traffic control for congested weaving areas and traffic incidents. The single-lane macroscopic traffic models which have commonly been applied for ATM purposes may fail to capture the complex traffic features on multiple lane roadways. These complex traffic features may consist of heterogeneous traffic flow distribution, capacity drop, and moving bottlenecks at different sections of the highway. Recently, research has revealed that vehicle lane-changing (LC) has significant impacts on road traffic safety since accidents often happen at lane-changing areas such as weaving sections and interchanges of the expressway. In view of this, there is a need to develop a comprehensive traffic model that can capture the effects of different lane-changing maneuvers explicitly on the surrounding traffic. One of the main objectives of this dissertation is to develop multilane traffic flow models to facilitate real-time simulation for various active traffic management applications. In the literature, the kinematic wave (KW) based approaches have commonly been used for simulating lane-changing maneuvers typically differentiate between mandatory lane-changing (MLC) and discretionary lane-changing (DLC). However, these two lane-changing behaviors were separately investigated by the existing traffic models. In view of this, a novel macroscopic multilane traffic model is firstly proposed in this dissertatoin to enable simultaneous simulation of MLC and DLC behaviors to capture the multilane traffic dynamics on freeway. A salient feature is that the proposed model does not require extensive traffic data collected by expensive infrastructure but only relies on the traffic data available to most of the traffic management centers. Such a parsimonious data requirement is a significant improvement over the existing traffic models. Other key features of the proposed model are: 1) incorporating the lane-based fundamental diagrams to encapulate the relationship between traffic speed-density and lane usage; and 2) modeling the drivers how to perceive the traffic condition spatially ahead to make their lane-changing decisions. To the knowledge of the author, these important features investigated in this dissertation for modeling vehicle lane-changing behaviors have not yet been reported in the literature. Apart from the above features, freeways are always subject to traffic demand and supply uncertainties, and noisy traffic data. To model the effects of these stochastic elements, a multilane traffic flow model is develped in this dissertation by extending the stochastic cell transmission model (SCTM) to simulate the effects of vehicle lane-changing maneuvers on freeway traffic dynamics. Link (cell)-node junction formulation is developed to propagate the lane-changing traffic. A fundamental speed-density relationship is used to interpolate the cell-lane speed profiles along a freeway corridor with sparse detectors. To the best of the author's knowledge, this is the first macroscopic stochastic multilane traffic model in the literature. The proposed models in this dissertation can be deployed as useful simulation tools for assessing the dynamic multilane traffic state based on the data available to the traffic management centers in practice. Furthermore, the model has the potential applications for predicting the impacts of various traffic incidents or lane control strategy on the expressway. In view of the advantages and the promising market prospect of the emerging connected automated vehicles (CAVs), the number of CAVs will be increased rapidly in the coming decade. Meanwhile, the regular human-piloted vehicles (RHVs) may still play a significant role in the roadway traffic. Therefore, it will be very likely that the roadway is to be shared by CAVs and RHVs in the near future. In the second part of the dissertation, an integrated optimal freeway traffic flow control framework that aims to minimize the total travel cost is devised for freeway traffic mixed with a given penetration rate of CAVs equipped with the Vehicle Automation and Communication Systems (VACS) and RHVs via en-route Variable Message Signs (VMS). It is assumed that the CAVs would follow full compliance with the control commands through the VACS. In contrast, the drivers of RHVs would make decisions in response to the information disseminated by the en-route Variable Message Signs (VMS). At the upper level of the integrated control framework, the objective is to devise an integrated action of several control strategies such as variable speed limit control (VSLC) and recommendation (VSLR), lane changing control (LCC) and recommendation (VSLR) under various traffic conditions. At the lower level of the integrated control framework, a multiclass multilane cell transmission model is developed to simulate the traffic flow dynamics mixed with CAVs and RHVs. The impacts of penetrated CAVs on the freeway traffic characteristics and the lane-changing behaviors are captured to design the optimal traffic control strategies. Firstly, the variations in the fundamental diagrams with respect to different penetration rates of CAVs are quantified. Then, the minimum headway acceptance criteria are determined for the lane changing (LC) maneuvers proposed by CAVs and RHVs with different motivations, respectively. An advanced priority incremental transfer (PIT) principle is adopted to evaluate the sending flows. Finally, the cell-lane-specific multiclass flow conservation law is developed to propagate the traffic flow and density on the freeway section. The effectiveness and the computational feasibility of the proposed optimal control strategies are illustrated via numerical example for a variety of penetration rates of CAVs under various traffic conditions. It is shown that the integrated control strategies can reduce the number of vehicles queuing at the bottleneck, improve the traffic efficiency and alleviate capacity drop. For road traffic safety, the integration of optimal control strategies can drastically reduce the instances of the stop and go traffic, smoothen the traffic flow and suppress the impact of the shockwaves on the freeway sections concerned.