Modelling Transport Demand in Sydney, Australia

Background

Transport for New South Wales (TfNSW) operates the Sydney Strategic Travel Model (STM) to inform long term transport planning, policy development, and infrastructure assessment in Greater Sydney, Australia.

The STM was originally developed between 1999 and 2002; during 2009 and 2010, the individual models that comprise the STM were re-estimated, and following that a number of projects have been undertaken to implement and further enhance the new models.

Research and Findings

Since 2009, RAND Europe have undertaken a series of projects to improve the STM:

Re-estimation of STM models. During 2009 and 2010 the STM models were re-estimated to use more recent survey data, the geographic scope of the model was extended as well as moving to a more detailed zoning system, and additional model components were incorporated to predict demand for toll roads and park-and-ride schemes.
Update the Population Synthesiser component of the STM. The Population Synthesiser is run to create forecasts of the future Sydney population, predict future levels of licence holding, and predict future car ownership levels.
Update the Application System component of the STM. The Application System comprises nine travel demand (TravDem) models; these were updated to use updated model parameters and to extend the TravDems so that they predict demand for the new toll road and park-and-ride alternatives.
Analyse and enhance the pivoting component of the STM. The pivoting component takes the predictions of the STM demand models in base and future years and generates the best possible predictions of future travel patterns. Some problems were identified with the original pivot process, though. RAND Europe made enhancements to the process to address these issues and undertook analyses to demonstrate the resulting improvement in performance.

While the overall performance of the new model was good compared to observed travel data, following testing of the new model TfNSW identified a number of other specific areas that could be improved.

Additional Estimation of the Sydney Strategic Travel Model thus describes further improvements to the commute mode-destination model, including an extension of the scope of the model to represent separate walk and bus access modes to train, changes to the model specification to predict higher walk usage in high density areas close to the Central Business District, and investigations of the impact of new terms for car driver. Changing the commute mode-destination model necessitated changes to other components of the forecasting system and these changes are documented in the report.

In late 2013, the model was revised to work with an average school day definition, rather than an average workday; this work is documented in STM3 modelling school days only. Furthermore, in 2014 the models were re-estimated to use a 2011 base year instead of a 2006 base year. This work is described in STM3 2011 base frequency, mode-destination and car ownership models.

Finally, in 2014 a study was undertaken that reviewed different studies that have considered how to model time period choice in strategic models and presented recommendations on how time period choice should be incorporated in the Sydney strategic travel model.