Imran, Waheed (2023) Macroscopic Modeling and Control of Mixed Traffic. [Tesi di dottorato]

Anteprima

Testo Imran_Waheed_36.pdf Download (44MB) | Anteprima

Abstract

Traffic congestion is more frequent on highways resulting from unprecedented urban development. Consequently, problems such as travel time delays, traffic pollution, and road traffic safety are major concerns for researchers. To diligently address the problem of traffic congestion, the dynamics of traffic flow on highways are vital to understanding for effective planning and management of highways. Effective traffic control strategies can be designed only if accurate traffic states are known. Thus, mathematical models of traffic are harnessed to predict the future states of traffic flow on highways. In this dissertation, the principal considerations center on the macroscopic continuum second-order traffic flow models. That being the case, these macroscopic models are employed in traffic control such as Model Predictive Control (MPC). Apart from traffic control, models of traffic are employed to describe the complex dynamics of evolving traffic and for various other purposes such as emissions estimations, incident impact assessments and traffic routing based on future dynamics of traffic flow. Lately, the macroscopic continuum traffic flow models are extensively refined for the characteristics of Human-Driven vehicles (HDVs), yielding positive findings, however, their practical utility is yet not fully exercised. Additionally, with the advent of Connected and Autonomous Vehicles (CAVs), certain dynamics of traffic flow such as their impact on the capacity of highways would be immense, and in particular, intrinsic characteristics such as driver behavior would significantly alter. Thus, it necessitates the need to revise the second-order traffic flow models for CAVs, and subsequently for CAVs and HDVs mixed traffic. In view of this rationale, the motivation of this dissertation is to evaluate the existing second-order traffic flow models, extend these models to mixed traffic, and later, investigate and harness them in traffic control, particularly, in MPC-based traffic flow control. To attain the goals of this dissertation, at the outset, a comprehensive review of the literature is conducted. Aiming to expand knowledge and understanding of the second-order continuum macroscopic models. These models are presented in distinct groups, which are categorically presented and discussed from a practical standpoint. A research gap in the current literature is identified considering the CAVs, which is addressed later in this dissertation. Subsequently, a detailed appraisal of the continuum second-order traffic flow models based on sensitivity analysis and numerical simulation is presented. The models are briefly studied and investigated to reproduce various traffic conditions. The second-order continuum models are examined considering their practical significance while analyzing the functional importance of the different parameters of the models. The sensitivity of the models' parameters is analyzed using statistical methods, signifying the understanding of revising them for mixed traffic. Further, a revamped model of mixed traffic is proposed. This model incorporates the distinct headway distance among vehicles CAVs and HDVs in mixed traffic. The proposed model is calibrated and validated on M50 Dublin motorway data reproduced in Simulation of Urban Mobility (SUMO). The proposed model is harnessed to investigate the impacts of CAVs on conventional traffic flow. While the additional practical significance of the proposed model is demonstrated by employing it in computing Level of Service (LOS) of the Dublin motorway segments with different Penetration Rate (PR) of CAVs. This model is based on a data-driven approach. Next, we proposed a novel continuum model for traffic mixture considering three classes of vehicles. To commence, the Fundamental Diagram (FD) is vital to understand, it is viewed as the basis of traffic flow theory in view of traffic mixture is not well studied. We delve deeper into the insights of the FD of traffic mixture and derive the FD of traffic mixture driven on class-specific varying headway distances of the vehicles. Subsequently, we propose a continuum second-order traffic flow model, aiming to effectively capture and analyze the traffic dynamics arising from the traffic mixture. Furthermore, the proposed model integrates the macroscopic emissions and fuel consumption model to quantify the exhaust emissions and fuel consumption resulting from the traffic mixture. Three traffic mixtures are considered (i) a mixture of CAVs and HDVs (ii) a mixture of CAVs, HDVs, and HVs, and (iii) a mixture of CAVs, HDVs, and CHVs. This model is based on a theoretical framework. Besides, a novel simulation-based approach is presented to evaluate traffic flow in the presence of CAVs. In this work, the anticipative behavior of CAVs is modeled based on the change occurrence time in traffic. Various dynamics of the macroscopic traffic flow are investigated providing the anticipative response of CAVs based on distance headway and density-dependent change occurrence time. Conclusively, an MPC-based Variable Speed Limit (VSL) control of motorway traffic is implemented by harnessing a continuum second-order traffic flow model for HDVs as a prediction model. Within this framework, congestion has been alleviated in some segments of the Dublin motorway, while on the other side, the congestion occurrence time is significantly minimized. Subsequently, we have used the same MPC framework to control traffic flow with CAVs, and we have explored alternative control variables instead of VSL in the case of CAVs. We use the time headway of CAVs as a control variable, and it yields sufficiently good results. In all, integrating the approach of modeling and control presented in this dissertation can help both the industry and academia in understanding modeling and controlling traffic flow in mixed traffic conditions.

Downloads

Actions (login required)

Modifica documento