Brief

Develop a proof-of-concept for the Simulation Framework and extend to the entire highway system in the Greater Toronto Area.

The SimFrame is a multipurpose, multilayer traffic simulation framework for the freeway and major arterial road network in the Greater Toronto Area (GTA). The MTO project team, which included engineering consultants Delcan Corporation and McCormick Rankin (MRC), developed a single master model that will not only consolidate local knowledge and best practices but also optimize modeling performance and efficiency. The intent of this new model is to become MTO Central Region’s primary tool for future traffic analysis as part of planning and design, operational reviews and traffic management strategy decision-making.

With the proof-of-concept completed, the goal is to expand the framework to encompass the entire highway system in the GTA and embark on large-scale, network-wide applications. This includes traffic planning and analysis of advanced traffic management strategies and a more comprehensive operational assessment of Central Region’s options for managed lanes such as high occupancy vehicle (HOV) or other priority facilities.

Method

Preliminary work has focused on urban portions of Highways 400, 401, 404, and 407; part of Ontario’s 400-Series controlled-access highway system. Segments of the 400-Series include various innovative traffic technology and systems such as HOV lanes, advanced traffic management systems, an electronically tolled highway and a collector–express roadway configuration in the Toronto area.

Modelling this huge and complex network is further complicated by the need to model heavy congestion and challenging physical features such as short sections between entry and exit ramps and transfer lanes between express and collector roadways.

Because of this combination of great size and dense detail, Delcan and MRC saw Aimsun’s integrated three tier macro-meso-micro approach as the best fit from among the nine proposals submitted as potential software platforms for the proof-of-concept. Aimsun has a single common network, a database that supports all levels of modeling, and a good interface with MTO’s existing Emme travel demand model. Aimsun also has proven consistency in modeling results from the micro and meso layers, plus powerful and efficient “zooming in” to focus on sub-areas for detailed analysis.

In parallel with the proof-of-concept effort and under the supervision of the MTO Traffic Office, Delcan and MRC have been putting the Aimsun platform through an extensive evaluation process since 2008. The testing is an ongoing process of assessment, calibration, validation, feedback and algorithmic enhancements and has included everything from the importation of existing traffic demand models to regional-scale evacuation of over a million vehicles. The consultants have checked that the framework is able to simulate every aspect of the 400 series network: reproducing queue lengths and shapes at ramps on multi-lane highways with peak flows; incorporating cost functions that allow faithful modeling of tollways and HOV lanes; and reproducing the process of traffic flow balancing between the express and collector lanes. In almost every case, feedback from these projects has led to important enhancements in the underlying software platform to tackle the complexity of Ontario’s highway network.

Conclusions

The SimFrame streamlines the simulation process as a centralized modeling framework and avoids a situation where the same road section appears in different models with potentially different modeling results. Consolidation also brings with it the benefits of standardized calibration, which assists in achieving consistency across studies. In addition, the SimFrame will lower costs by moving away from single-use models to a model that can be re-used time and time again. Finally, starting from an existing base model means swifter results as it is no longer necessary to build and calibrate a new model from scratch for each new project.

Wide-area projects that cannot be modeled efficiently at the microscopic level due to the large size of the study area, but that require more detail than that provided at the macroscopic level, can now be addressed using the mesoscopic level. Furthermore, with SimFrame, there is no need to maintain and update separate models with independent networks and databases.