Decentralized transport management: how can simulation solutions pave the way?

How can we estimate the efficiency margins of new decentralized transport management schemes and evaluate their potential to increase sustainability? This is how Aimsun has been facilitating the delivery of new simulation-based scheme analysis frameworks within the H2020 DIT4TraM project.

Why do we need to decentralize transport management?

Transportation ecosystems are constantly changing. Societal developments, new technological trends, and the emergence of disruptive mobility services have resulted in systems with many interacting agents – system stakeholders like travellers, connected vehicles, service providers, policy makers, road authorities – each adapting to their perceived past, present, or future situations on different timescales.

Traditional approaches to transportation systems relied on centralized traffic control and network management schemes. Centralized paradigms may only have limited success due to i) computational complexity and high problem dimensionality, and ii) the lack of coordination and regulation of information exchange between different system components and agents.

Adding to the “equation” the increasing complexity stemming from the multi-layered interactions of emerging transport systems with new services and technologies renders centralized optimization of transportation systems even more challenging.?

Paving the way to higher system efficiency and sustainability via decentralization – the DIT4TraM approach

To address this problem and, at the same time, support the transition to seamless and sustainable, connected, and autonomous mobility, the H2020-funded DIT4TraM project adopts, investigates, and eventually applies new paradigms for smart, holistic, and unifying transport management.

Through the concepts of decentralization, co-operation, and negotiation, DIT4TraM investigates and builds on existing distributed (swarm) intelligence theory towards disentangling transport systems’ components as much as possible, while ensuring sufficient system-wide cooperation and coordination via smart system designs to achieve societal optimum states. The resulting generic distributed control principles constitute the foundations for the formulation of new decentralized transport management schemes and optimization frameworks on four different application domains, i.e., i) traffic operations, ii) mobility management, iii) demand-supply synchronization and iv) shared mobility services.

To test the efficiency gains of these novel schemes, in either real-life or virtual (simulation) contexts, DIT4TraM is establishing and developing different prototypes of decentralized multi-agent systems in six different European pilot areas. Within several of these systems lies a novel Distributed Network mobility and Traffic Management (DNTM) platform – a common and reusable (among systems) communication and orchestration platform acting as a system facilitator with a twofold functionality. The first relates with real-life system deployment features – i.e., enabling the deployment of the project’s decentralized schemes and optimization frameworks in real-life dynamic settings as well as the communication between the system’s agents (e.g., traffic lights) and their environment (e.g., censors). The second relates with scheme analysis and evaluation features for planning – i.e., enabling system’s stakeholders (e.g., traffic managers, infrastructure managers, planning authorities) to analyse, test and evaluate those strategies and their system-wide impact at pre-deployment stages using simulation-based scheme analysis frameworks.

In DIT4TraM, Aimsun is facilitating and enabling the development of such frameworks by integrating our state-of-the-art modelling and simulation solution, Aimsun Next, with DIT4TraM’s DNTM platform and its scheme optimization controllers. The integrated DNTM platform-Aimsun Next framework will be set up and applied for analysing and evaluating the efficiency and impact of different design scenarios for decentralized management schemes, including i) model-predictive traffic signal control, ii) model-free (using Machine Learning) perimeter traffic control and iii) highway demand management via Tradable Credit Schemes (TCS). In this article, we focus on the TCS simulation-based analysis framework, and we elaborate on the preliminary integration, modelling, and simulation approaches. Furthermore, we discuss the project’s Mediterranean Highway (AP-7) pilot.

Tradable credit schemes for interurban traffic management

What is a tradable credit scheme, or TCS? ?Tradable mobility credits are a new alternative to conventional congestion charging schemes. In TCS, travellers receive an initial allocation of credits that they can consume when travelling by car or other transport modes. The credits can also be traded - the decentralization element – i.e., heavy users can buy extra credits, while light users can sell credits. This new concept allows policymakers to directly address equity issues by providing different credit allocations based on socio-economic variables. It also allows for an ‘internalization of externalities’, i.e., the value of a credit can be directly equated to a measurable unit, such as CO2 emissions. Since externalities in transportation mostly depend on mode, time, and speed, authorities can use the mobility credits to encourage sustainable travel. ?Detailed information on TCS can be found in DI4TraM’s Deliverable 4.1: “Tradable Mobility Credits and Permits: state-of-the-art and concepts”.

In DIT4TraM, both urban and interurban TCS applications are investigated to capture the degree of sustainable impact the scheme may have on different scales via influencing travellers’ or drivers’ travel behaviour – either on day-to-day level travel patterns (e.g., mode switch, activity participation) or on more within-day aspects like routing. However, here we focus on the interurban case, and we elaborate further on the context and framework of a virtual investigation (simulation-based analysis) for a potential TCS application on a motorway in Spain.

More specifically, since January 2020 and September 2021, the Spanish government and highway management authorities have decided to (temporarily) remove the tolls from several Spanish motorways, including parts of the AP-7 (Autopista del Mediterráneo). This has resulted in increased traffic and lower Levels of Service (LoS) across several parts of the Spanish motorway system, particularly the AP-7. As such, in DIT4TraM we investigate TCS under both regular and irregular conditions and assess the potential of replacing the previous tolling system with a more equitable means of traffic management towards achieving three goals:

• Alleviate habitually congested and low Level of Service (LoS) arterial corridors (highways) by balancing demand among different highway alternatives (day-to-day scope),
• Reducing emissions caused by interurban travel,
• Diffuse congested corridors resulting from emergency situations, e.g., accidents (within-day scope)

The highway network of the Catalan region of Girona has been selected as our virtual testbed for the virtual TCS application at hand. The figure below illustrates the scope of the Girona highway model for the TCS application (right) and the scope of Aimsun’s existing microscopic AP-7 model (left).?The scope of the Mediterranean highway model used by the TCS analysis and assessment framework includes two highways, i.e., two route alternatives, in the periphery of the Girona metropolitan area – AP-7 and N-II. AP-7 is a Spanish highway running across the Northern and Southern borders of Spain, while N-II is a national motorway connecting Madrid with La Jonquera (a city at the French border). The exiting microscopic AP-7 model in Aimsun Next has been calibrated with data from 2019 (pre-pandemic and pre-toll-removal) and will be further extended (addition of N-II infrastructure) and re-calibrated with openly available traffic flow data[1] from Catalan and Spanish authorities.

Due to the lack of post-pandemic traffic flow (counts) data (due to the toll removal), other factors will be used to adjust the highway demand profiles to post-pandemic "recovery" conditions based on either sensitivity analysis approaches or using aggregate total statistics from available reports on highway performance. The developed microscopic highway model in Aimsun Next will be used to simulate traffic running across the northern and southern intersections of the two highways (North and South of Girona), assuming different penetration rates for vehicle enrolment to TCSs with different “designs” (credit allocation, credit charges for alternatives, etc).

An agent-based traffic microsimulation framework for interurban TCS analyses via the DNTM platform

The simulation-based analysis framework that will be used for the TCS assessment in DIT4TraM’s Mediterranean Highway pilot is the result of the interoperable coupling among a TCS ecosystem module, developed by the LICIT-ECO7 laboratory at University Gustave Eiffel, and the Aimsun Next microsimulation solution. The conceptual architecture of the integrated modelling and simulation framework is illustrated in the figure below.

The TCS ecosystem module is a SUE-based (stochastic user equilibrium) modelling framework that estimates and captures the main dynamics, i.e., actions and behaviours, of stakeholders involved in the TCS and their interdependencies. More specifically, it i) captures users’ (travellers’) choices (route choices between AP-7 and N-II in this context) via a logit-based model that is coupled with the credit marketplace, ii) it estimates credit prices arising from credit marketplace dynamics and iii) defines the credit charging profiles resulting from a regulating authority’s actions towards satisfying an array of societal objectives (e.g., total travel time, total carbon emissions).

The basic notion governing the module’s functionality is that the regulating authority (e.g., highway infrastructure manager) rations access to the AP-7 highway - as credits are needed to drive on it – by computing the optimal credit charges. An initial credit allocation is provided to the travellers (drivers), and they can trade them on the dedicated credit market. Based on the resulted credit charges, the credit prices (from the credit marketplace) and expected travel times (from the Aimsun Next simulator), TCS users are making route choices (AP-7 or N-II), which are then simulated in the microscopic Mediterranean highway model of Aimsun Next.

The proposed simulation-based analysis framework enables:

• elicitation of disaggregate demand from calibrated microscopic models (input demand required by TCS), and
• external (from the DTA-based microsimulator) route assignment considering TCS-specific “costs” (input demand, i.e., vehicle trips, required by the Next simulator)

Mode-specific travel demand information, i.e., disaggregate travellers (multi-class vehicles) with specific trip Origins and Destinations, is one of the main (offline) inputs required by the adjusted for this use-case TCS module. Given pilot- and scenario-specific limitations, i.e., population and demand data availability for interurban scope, eliciting disaggregate highway demand will rely on traffic counts extracted from the calibrated microscopic highway model in Aimsun Next and converted into synthetic agent (vehicle) trips. Due to the lack of VoT (Value of Time) information for this scale and scope as well as pilot-specific survey data indicating travel (mode or route) attitudes in the presence of a TCS, route choice models from the literature will be adopted along with a sensitivity analysis for the main parameters affecting traveller’s route choices in TCS.

Due to the disaggregate (agent-based) nature of demand representation in the TCS ecosystem module, the Traffic Arrival traffic simulation feature of Aimsun Next will be used to simulate agent (vehicle) trips with specific routes. A Traffic Arrival object can be generated by modellers to create a specific, and customized, pattern of arrivals of vehicles into a simulation at a specific time and place (from-to). Traffic arrivals can be generated externally from other sources other than Aimsun Next via an XML file which is then linked to a Traffic Arrival object in the same way as if it were to be generated by a simulation replication.

To enable end-users of DIT4TraM’s DNTM platform to utilize this novel simulation-based TCS analysis framework after the DIT4TraM project’s completion, the proposed framework is being integrated into the DNTM platform via Aimsun Next’s flexible API design and a Central Broker from the DNTM system. The integration (both time-based and event-based) will be achieved via two different interfacing workflows according to the simulation and contextual particularities of each simulation use-case study. In the day-to-day case (achieving load balancing across the highway infrastructure network in the mid-term), the interaction occurs only once per iteration of a daily scenario. In this case, pre-trip information, i.e., vehicle routes, are computed by the TCS module a priori and such information is communicated with Aimsun Next only once and before the simulation starts (scenario’s demand).

In a within-day process for handling emergencies and unplanned events, the coupling occurs in both a time-based and event-based manner. In the time-based case, the simulator stops at a fixed time-step interval (e.g., every 5 mins), and sends information (e.g., travel times, Levels of Service (LoS)) to the TCS module to permit the regulator to adjust the credit charge dynamically. The regulator adapts the credit charge in the mid-term to reach the optimum for a certain criterion. In the short-term and to react to unexpected events, the event-based interfacing approach will enable communication of information in case specific events (e.g., sudden decrease in LoS, increasing travel time in a highway, increasing congestion) take place in the simulation. For example, the regulator may increase the number of credits needed, and thus the travel cost of the highway, to mitigate the effect of an accident on the AP7 (or N-II) on short notice.

?

The value of simulation in testing new management concepts

Simulation-based analysis frameworks are key tools for transportation authorities, operators, and infrastructure managers in the process of decarbonizing transportation via new decentralized concepts. Understanding the efficiency margins from such management schemes and how different scheme designs may impact multi-agent ecosystems and their environments necessitates the integration of new technological and system dynamics in simulation environments. We are looking forward to bringing you more insights from the TCS application in the Mediterranean Highway pilot of DIT4TraM very soon.

[1] https://www.mitma.es/carreteras/trafico-velocidades-y-accidentes-mapa-estimacion-y-evolucion/mapas-de-trafico/2019