51³Ô¹Ïapp

With the advent of urban sprawl, managing congestion is at the heart of major cities’ operational challenges. The prediction of travel time variation on congested corridors in urban and peri-urban networks is critical to inform travellers of traffic conditions and assist them in their trip planning. Travel time predictions are also extremely relevant to forecast and assess network performance and anticipate congestion episodes through the use of variable message signs. Further road capacity reductions caused by incidents, road occupancy projects (e.g. maintenance operations) or special events may also impact travel time and should be accounted for in the travel time forecasting process.

The focus of Application 1 was to develop a real-time travel time prediction software to provide an effective and robust framework for predicting travel time during peak and off-peak periods at a corridor-level. The developed travel time prediction software innovates on the available forecasting tools by allowing dynamic road capacity variations to be tracked in real-time and providing adapted travel time estimations. This is possible thanks to an innovative modelling framework which incorporates traffic flow dynamics within the travel time forecasting process instead on relying exclusively on historical data. This project is envisioned to contribute towards a framework for intelligent transportation systems.

Ìý

To study traffic dynamics at a network level, traffic flow should be evaluated from a fine resolution. Considering the ever changing traffic conditions in a city, Dynamic Traffic Assignment (DTA) models are useful to access the impact of policy measures on the travel behaviour and overall network performance at a regional level. A majority of established planning models in Australia (and around the world) are macroscopic in nature and provide aggregated performance measures for the entire network. However, these traditional models cannot represent important phenomena such as queue spillback or temporal congestion propagation due to their time-invariant framework. One solution to this issue that is receiving greater attention in both research and industry is DTA modelling.

Application 2 is focused on developing and deploying a metropolitan area dynamic assignment model (MADAM) for the great metropolitan Sydney region. The workflow for the project involved obtaining the necessary input data, including network geometry, travel demand, signals and transit, converting the data into the form needed for the DTA platform, making model adjustments for computational efficiency, implementing and finally calibrating the model. This pioneer project helped to understand the role of dynamic traffic assignment models thanks to many lessons learned during deployment and calibration process, and thanks to extensive results analysis. This study makes a significant contribution toward developing a regional dynamic model for a metropolitan city in Australia. In the future, the MADAM model could aid in evaluating important policy decisions and infrastructural development in the context of the overall network operation. This project serves as a proof of concept and provides valuable insights to other practitioners interested in the areas of transport planning and traffic modelling.

Ìý

Building on Application 1, Application 3 extends the real time travel-time prediction framework at the sub-network level, accounting for multiple origins and destinations. This aims to identify route choice strategies when several paths between an origin and a destination are possible. This project explores existing route choice models to identify which approach is appropriate for the prediction of travel times over a sub-network. A tailored route choice model has been developed based on empirical observations and integrated within a network-level travel time prediction framework. The proposed methodology is expected to be able to produce information that can evaluate planning scenarios, aid travellers through information provision and assist in real-time management of traffic conditions (including incident management).

Travellers using the transport network will be the natural users of this research since the real-time travel time prediction software can be used to provide effective and robust estimations of travel time during trip planning and forecasting. The incorporation of road capacity reductions within the travel time forecasting process will further emphasize the importance of the research by providing travellers with adaptive real-time travel time information which is currently seldom available. In addition, this offers the possibility to anticipate congestion episodes caused by road occupancy projects and special events, which are very frequent in large metropolitan areas. Finally, the possibility to run what-if scenarios corresponding to multiple road capacity reductions will allow traffic management centres to better absorb network disruption and the forecasted increase in travel demand.

Ìý

Dynamic traffic assignment (DTA) research has advanced greatly in terms of deployability, computational feasibility, and representing complex temporal phenomena. There have also been substantial contributions regarding various aspects of stochasticity within DTA. However, there are persisting limitations in terms of approaches which are both computationally tractable and provide more detailed representation of stochastic aspects.

Application 4 explores the application of a novel Strategic User Equilibrium DTA (StrDTA) modelling framework, which captures the impact of users making a priori route choice decisions based on the knowledge of a range of possible demand scenarios (e.g., differing days or representative situations). The resulting stochastic DTA problem becomes complex due to the integration of multiple demand scenarios and the algorithmic adjustments necessary to find optimal paths. The proposed new solution approach was implemented, and a detailed case study for the Sydney Central Business District (CBD) network was conducted. Results demonstrate the importance of accounting for stochasticity in the routing algorithm rather than relying on assumptions of average values. Similarly to previous applications of the project, Application 4 not only focuses on delivering cutting-edge research but also targets existing practical modelling questions and provides insights especially valuable to practitioners active in the areas of both traffic modelling and transport planning

Ìý