Gold Coast: Predictive Solutions Trial

Brief

Queensland’s transport system is made up of interacting elements such as motorways, arterial roads, tunnels, vehicles, and public transport services, which are monitored and managed to ensure the system works optimally, with minimal congestion and delays avoided where possible.

An area of interest for the Department of Transport and Main Roads (the department) is the transition from reactive to proactive network management, through the use of technologies such as big data and machine learning to support better transport operations. The department maintains a number of existing applications and business processes that have the potential to be impacted by such technology.

Through 2016/17, the Predictive Solutions trial delivered an operational implementation of the Aimsun Live solution. This trial aimed to assess the accuracy of Aimsun Live’s predictions and simulations, the potential limitations of existing technologies and data sources, and the implications for the department such as maintenance
requirements and its likely value to operations. The goal is to enable the department to better predict likely issues and determine the best strategies to prevent or mitigate issues on the transport network, instead of reacting when they occur.

Background

The Aimsun Live trial is part of a broader program of works which focus on investigating opportunities to better use technology and data to get more out of the existing transport network. This includes opportunities relating to network optimisation, incident management, traveller information, and transport coordination.

Project Scope

The trial ran during weekday morning and afternoon peak-hour periods across a small area on the Gold Coast, and used live data inputs from the STREAMS ITS (Intelligent Transport System) platform, including the SIMS incident management system.

The study area was based on part of an existing model, and covered about 20% of the Gold Coast metropolitan area. The trial period ran from August 2016 to June 2017.

As this was a trial the solution was built and executed online, however there was no change to existing operational systems during this period, and no integration into operations. It was acknowledged that this presented limitations on the output of the solution, however this was assessed and recorded as part of the trial.

Conclusions

Despite limitations in the trial in terms of geographical area, time and technology, it provided significant insight into the opportunity presented by simulation based decision support tools, and the necessary business and system changes that would be required to achieve the best outcomes.

The lessons learnt from this trial will inform future phases, such as implementing a 24 hour monitoring/prediction system, and defining a perimeter area including all the observable perturbations of the response plans applied in the network.

The quantity and quality of data inputs had a large influence on the final results, and on the quality of the predictions and evaluations of the future traffic conditions. These data inputs include both network layout, performance, as well as incidents. For future phases, there would need to be greater focus on availability and accuracy of traffic data collection and definition (including detectors/stations, reliability of the data, quantity of data available), and on the incident feed used to generate events to model into the simulation network.

Despite the limitations presented to the trial, the insights gained were of great value, and are being used to inform further departmental projects.