The Macroscopic Fundamental Diagram (MFD) describes the relation of average network flow, density and speed in urban networks. It can be estimated based on empirical or simulation data, or approximated analytically. Two main analytical approximation methods to derive the MFD for arterial roads and urban networks exist at the moment. These are the method of cuts (MoC) and related approaches, as well as the stochastic approximation (SA). This paper systematically evaluates these methods including their most recent advancements for the case of an urban arterial MFD. Both approaches are evaluated based on a traffic data set for a segment of an arterial in the city of Munich, Germany. This data set includes loop detector and signal data for a typical working day. It is found that the deterministic MoC finds a more accurate upper bound for the MFD for the studied case. The estimation error of the stochastic method is about three times higher than the one of the deterministic method. However, the SA outperforms the MoC in approximating the free-flow branch of the MFD. The analysis of the discrepancies between the empirical and the analytical MFDs includes an investigation of the measurement bias and an in-depth sensitivity study of signal control and public transport operation related input parameters. This study is conducted as a Monte-Carlo-Simulation based on a Latin Hypercube sampling. Interestingly, it is found that applying the MoC for a high number of feasible green-to-cycle ratios predicts the empirical MFD well. Overall, it is concluded that the availability of signal data can improve the analytical approximation of the MFD even for a highly inhomogeneous arterial.

This paper describes a general approach to the specification the stable regime speed-flow function, for motorways, as a part of the stable regime Stochastic Fundamental Diagram consistent with main assumptions of Transportation Systems Theory. Main original elements are: • Specification of speed-flow functions consistent with travel time function, such as BPR-like functions; • Calibration from disaggregate data, say data from single vehicle trajectories; • Specification of the speed r. v. distribution consistent with those used in RUT for route choice behavior modelling, such as Gamma, Inv-Gamma.

Given the increasing consciousness toward the environmental footprint of mobility, accommodating environmental objectives in existing transport planning strategies is imperative for research and practice. In this paper, we use the link macroscopic fundamental diagram (MFD) model to develop optimal routing strategies that minimize total system emissions (TSE) in multiple origin-destination (OD) networks. Piecewise linear (PWL) functions are used to approximate MFD for individual links, and to define link-level emissions. Dynamic network constraints, non-vehicle holding constraints, and convex formulations of the PWL functions are considered. Thus, the system-optimum dynamic traffic assignment (SO-DTA) problem with environmental objectives is formulated as a mixed integer linear program (MILP). Finally, on a synthetic network, numerical examples demonstrate the performance of the proposed framework.

Being widely adopted by the transportation and planning practitioners, the fundamental diagram (FD) is the primary tool used to relate the key macroscopic traffic variables of speed, flow, and density. We empirically analyze the relation between vehicular space-mean speeds and flows given different signal settings and postulate a parsimonious parametric function form of the traditional FD where its function parameters are explicitly modeled as a function of the signal plan factors. We validate the proposed formulation using data from signalized urban road segments in Salt Lake City, Utah, USA. The proposed formulation builds our understanding of how changes to signal settings impact the FDs, and more generally the congestion patterns, of signalized urban segments.

La congestion en milieu urbain est un enjeu majeur que ce soit d’un point vue économique, social ou environnemental. À court et moyen terme, l’utilisation de la simulation dynamique du trafic routier peut permettre d’analyser et de guider des politiques d’optimisation des infrastructures existantes. Aujourd’hui, du fait de la complexité des systèmes de transport, les outils de modélisation classiques sont limités à des échelles géographiques peu étendues (de l’ordre du quartier). À grande échelle, le temps de calcul devient rapidement un facteur limitant tout comme le calibrage et la scénarisation. Néanmoins les dernières décennies ont vu l’apparition d’une nouvelle génération de modèles bien adaptés aux métropoles urbaines. Ceux-ci sont basés sur une relation phénoménologique entre la production de déplacements et le nombre de véhicules dans une zone spatiale d’un réseau routier, appelée Diagramme Fondamental de Zone (Macroscopic Fundamental Diagram, MFD). Cette relation, validée empiriquement sur de nombreuses villes, a permis d’étudier différentes méthodes de contrôle du trafic pour une ville entière, mais a été peu utilisée à des fins de prévision de la congestion. L’objectif de cette thèse est de proposer un premier outil opérationnel de simulation et d’analyse des grands réseaux de métropoles, en utilisant et développant les modèles de trafic basés sur la relation MFD. Cet outil doit posséder un cadre théorique cohérent qui puisse convenir à des applications telles que la prévision d’états de trafic, le développement de nouvelles politiques de contrôle, l’estimation de pollutions liées au trafic, etc. Les contributions de la thèse portent sur deux aspects. Le premier est l’analyse des propriétés mathématiques et physiques des modèles existants, en incluant une formalisation complète de la gestion de plusieurs longueurs de parcours au sein d’une même zone urbaine. En particulier, cette formalisation traite de la distinction des trajets internes à la zone et des problèmes de flux convergents et divergents pour les trajets traversant la zone lorsque la congestion se propage d’une zone à l’autre. Le deuxième aspect est la proposition d’un nouveau modèle basé sur la distance individuelle parcourue à l’intérieur d’une zone urbaine (trip-based). Cette approche permet d’individualiser les usagers (auparavant représentés sous forme de flux continus) et donc de définir plus finement leurs caractéristiques, en vue de coupler leurs déplacements à des modèles d’affectations sur différentes routes. Enfin, des exemples d’application illustrant diverses collaborations sont donnés en dernière partie de la thèse. La simulation du trafic sur l’aire urbaine du Grand Lyon (France) y est présentée, ainsi que de nouveaux modules de modélisation de la recherche de parking ou de contrôle périphérique. Cette thèse est partie intégrante d’un projet européen ERC intitulé MAGnUM : Approche multi-échelle et multimodale de la modélisation du trafic pour une gestion durable de la mobilité urbaine. / Congestion in urban areas has become a major issue in terms of economic, social or environmental impact. For short or mid term, using dynamic road traffic simulation can help analyzing and providing guidelines to optimization policies of existing infrastructures. Today, because of the complexity of transport systems, classical modeling tools are limited to small geographical areas (of a district size). Computational time, together with simulation calibration, are notably very constraining at large scales. However, a new generation of models designed for metropolitan areas has arisen over the past decades. These models are based on a phenomenological relationship between travel production and the number of vehicles in a given spatial area of a road network, known as the Macroscopic Fundamental Diagram (MFD). This relationship, supported by empirical evidences from several cities around the world, has allowed the study of different traffic control schemes at a whole city scale, but was rarely used for traffic state forecasting. The aim of this PhD is to propose an efficient modeling tool, based upon the concept of MFD, to simulate and analyze traffic states in large metropolitan areas. The theoretical framework of this tool must be consistent and applicable for traffic state forecasting, development of new control policies, traffic emission estimation, etc. There are two major contributions in this PhD. The first one is analyzing the mathematical and physical properties of existing models, and formalizing the dynamics of several trip lengths inside the same urban zone. In particular, this formalization distinguishes between internal trips and trips crossing the zone. Flow merging and diverging issues are also addressed when congestion propagates from one zone to another. The second contribution is proposing a new trip-based model based on individual traveled distance. This approach allows to treat users independently (previously represented with continuous flows), and thus to define their characteristics more precisely to couple their trips with assignment models on different paths. Finally, examples of application from various collaborations are given in the last part of this thesis. It includes a simulation study of the Grand Lyon urban area (France), as well as new modules to simulate search-for-parking or perimeter control. This PhD is part of a European ERC project entitled MAGnUM: Multiscale and Multimodal Traffic Modeling Approach for Sustainable Management of Urban Mobility.