DynusT (Dynamic Urban Systems for Transportation) is a simulation-based dynamic traffic assignment (DTA) model with a 15 years development history that can support engineers and planners to address emerging transportation planning and operation challenges. DynusT users can estimate the evolution of system-wide traffic flow dynamics patterns (resulting from individual drivers seeking the best routes to their destinations) and respond to changing network demand, supply, or control conditions.
These application areas for DynusT include assessing the impacts of alternate traffic operations and control strategies, and evaluating strategies for work zone, incident, and special event management. Engineers or planners can also use the software to assess the impacts of intelligent transportation system technologies, such as dynamic message signs, ramp meters, and in-vehicle guidance systems. In addition, DynusT can evaluate congestion-pricing schemes for toll roads and produce traffic operation data for air quality analyses.
DynusT is developed based on decades of rigorous theoretical developments and practical applications by the research team led by Professor Yi-Chang Chiu at the University of Arizona. The software is now optimized and QA/QCed by Metropia's software team.
A robust, practical, and trustworthy DTA model needs to simulate traffic flows that exhibit desirable flow properties. Various traffic controls need to be adequately represented, and various driver travel choices need to be accounted for. The simulation and assignment algorithm needs to be computationally efficient and the dynamic user equilibrium (DUE) convergence condition needs to be truthfully sought for. The model also needs to be easy to calibrate so that the users can finish the project with desirable outcomes within the budget and schedule. Here is an overview of DynusT.
TRAFFIC/USER BEHAVIOR SIMULATION
Anisotropic Mesoscopic Simulation (AMS) that exhibits micro-like traffic flow properties at 1000x the speed of micro-models. AMS combines the best of both worlds - micro car-following properties and computational efficiency. More technical details about AMS can be found here.
Up to 10 traveler classes to allow flexible pricing/roadway access control that may exist in the real world. These 10 classes were fully tested in the recent activity-based model (ABM) and DTA integration project commissioned by FHWA for the Atlanta Regional Council (ARC).
Vehicles in DynusT can follow one of the four routing objectives. The pre-trip objective is to follow the pre-defined/historical path without learning a new route. The DUE objective is to continually learn and switch to a better-experienced route until no improvement can be made. The pre-trip optimal routing follows the best route at the time of departure (such as following the route recommended by the navigation app). The en-route rerouting allows the route to be regularly updated during the journey (such as the WAZE type of routing). DynusT allows a model user to specify a certain mix of traffic following these four objectives to reflect the actual routing behaviors in a real-world situation.
Five types of intersection controls - pre-timed, pre-timed coordinated, actuated signal controls, stop signs and yield signs - can be modeled in the simulation. This simulation logic accurately captures the arterial traffic conditions.
Other ITS systems such as Dynamic Message Signs (DMS) or freeway control like ramp meters are also modeled in DynusT.
Modeling work zones is rather flexible in DynusT. Once the model user specifies the location of the work zone, capacity reduction, and the work zone speed limit (optional) DynusT can accurately represent the merging and traffic flow shock waves caused by the work zone induced bottleneck.
Simulation is highly paralleled (OpenMP in shared memory) in computation to take advantage of the model computer parallel processing architecture.
DynusT permits a rather flexible way of generating vehicles: dynamic OD matrices only, trip rosters only, or the hybrid loading of OD matrices and trip roster. One can find this flexibility rather useful in several scenario management situations discussed below. The integration with the activity-based models (ABM) requires the hybrid loading of trip rosters (commuters) and OD matrices (external freight traffic or airport trips).
The time-varying OD matrices can be specified in a wide range of time resolutions from minutes to hours.
DynuStudio allows a model user to effectively manage multiple scenarios in which both demand and supply conditions may be modified. A model user may use the OD matrices to generate trip rosters in the baseline scenario, then use the DynuStudio's scenario manager to load the same trip roster across all other scenarios in order to fix the demand while evaluating the system performance difference due to the modeled scenario.
DynuStudio also allows the model user to compare two scenarios graphically and in animation in order to have a visual contrast to the compared scenarios.
DYNAMIC USER EQUILIBRIUM
DynusT employs a highly computationally efficient, parallel processed (OpenMP in shared memory), time-dependent, shortest-path algorithm with temporal domain partitioning techniques to be able to perform traffic assignment with minimal run-time and memory requirements. This is the primary reason why DynusT is the only DTA model that has the most successful large-scale model case studies.
MODELING PRICING WITH INDIVIDUAL VALUE OF TIME
DynusT is the only DTA model which has proven to be able to computationally efficiently perform DUE with each vehicle carrying its own value of time. This feature is the result of the methodological and algorithmic breakthrough by the DynusT research team in the recent years. This allows the pricing related studies to be conducted in a much more realistic manner. An after-implementation back-casting of a recent pricing study in Denver in 2016 showed DynusT's traffic and revenue results are highly comparable with actual real-life observations.
For more publications about the theories of DynusT or completed projects, please click here.
For more information about the DynusT overview, tutorial, or project videos, please click here.
For recent published research reports based on DynusT, please click here.