Rethinking strategic transport planning practices in SA

Travel demand models are an essential part of transport planning and policymaking. They’re used by metropolitan, regional, and national planning organisations to devise the capacity for new infrastructure. Models help balance economic, environmental, and social factors thereby developing credible and reliable forecasts.

The project, The future of travel demand modelling and forecasting, was funded by the iMOVE Co-operative Research Centre and the South Australian Department for Infrastructure and Transport (DIT). It was carried out by The Institute for Choice (I4C), based at the University of South Australia, which aims to identify future research directions.

What follow here is a wrap-up of that project, and at the bottom of the article, a downloadable copy of a wrap-up report, Rethinking strategic transport planning practices in South Australia.

Objectives

The study’s two objectives were to:

• Identify the department’s strategic transport planning needs
• Determine appropriate decision-support systems that could support those needs.

What linked these objectives was a focus on current and future disruptions to the transport sector. These included shared, electric, connected, AI, and autonomous transport modes.

Travel demand models (and why new models are needed)

Travel demand models are quantitative tools that offer insights into current travel behaviour patterns and a framework to produce behavioural change in response to one or more future transport investments. They produce forecasts that help regional planning bodies:

• Determine the required capacity for new infrastructure needs, and
• Assess economic, environmental, and social impacts that organisations must consider in funding competing initiatives.

However, travel demand models can be unreliable in their absolute predictions. For instance, it’s difficult to forecast human behaviour several years hence. But such models are stronger when they compare the expected relative impacts of different projects. For example, the South Australian Government was expected to spend $11.3 billion in the four years to 2023 to develop its infrastructure.

Typically, planning organisations use traditional four-step trip-based models or more modern activity-based models. Tour-based models are rarely used, particularly because they are not considered as behaviourally realistic as activity-based models. As well, a tour-based model will not capture spatial and temporal constraints impact multiple ‘tours’ a survey respondent takes in one day.

However, the researchers were mindful of the department’s resource constraints which precluded the use of activity-based models. That’s why the project recommended using a couple of models one being existing four-step trip-based models to underpin the analysis.

This approach segments the population based on demographic and spatial characteristics. The concept draws on household characteristics such as household size, car ownership level, number of employed household members and dependents. It also assumes all the households have similar preferences. It is known as a deterministic approach but has some probabilistic elements. This approach may not be able to capture the full complexity and variability of human travel behaviour, though.

Another approach that could work is a probabilistic segmentation process, such as the latent class models. They identify segments or classes with preference differences. This model can incorporate more variables than using a primarily deterministic approach.

Roadmap for improvement

The project acknowledged the department is no longer solely responsible for building transport infrastructure and delivering transport services. Instead, these responsibilities have increasingly been moved to other public sector departments and outsourced to the private sector. The shift sees the Department of Planning and Transport Infrastructure taking more of a transport policy leader role.

To bring transport demand models up to speed, the researchers created a roadmap with four milestones for the department:

Short-term:

• Boost organisational practices for the model’s governance and ownership. Improve engagement with end-users.

Short-to-medium-term:

• Use new data sources to recalibrate and validate existing and potentially new models

Medium-to-long term:

• Identify how the existing model structure should be changed so it can support long-term strategic planning

Long-term:

• Review the South Australian government’s long-term vision for the state.
• Consider how transport and land use fit into the picture.
• Determine how transport modelling through detailed scenario planning can support that vision.

Methodology

Researchers conducted an extensive literature review and interviews. They canvassed the views of stakeholders in government, industry, and academia as well as grey literature (non-scholarly/peer-reviewed research reports or documents).

This comprehensive approach spanned several decades from the past and into the future, and included interviews with:

• The department’s senior leaders and other key staff about its past practices and future needs
• National and international transport planning organisations to compare with the department’s perspective
• Relevant national and international academic experts to pinpoint innovative research underway, and
• A review of grey literature on planning process and the decision-support systems underpinning them.

This methodological approach was not a review of national and international best practices about travel demand modelling and forecasting because other scholars have already published excellent reviews of the literature.

Report findings

Three key findings and recommendations emerged:

1. Use the strategic models to compare the expected relative impact of different projects and to prioritise major transport investments
2. Adopt scenarios in the assessment process to account for future uncertainty and focus on investments that optimise resilience across several future possibilities, and
3. Urgently re-calibrate and validate with current data the strategic models that could be used to support this assessment process within the department.

Importantly, the report found the department’s limited resources undermined the performance of the current transport demand models. For instance, the strategic model for Adelaide’s metropolis was last recalibrated using 1999 household survey data. That approach no longer matches current or future traffic patterns within the region. Back in 1999, EVs, shared mobility services, connected and autonomous vehicles weren’t on the radar. As well, consumer preferences have significantly changed, including declines in:

• Car ownership
• Licensing rates
• Private car use, and
• Demand for suburban neighbourhoods as inner-city areas gain in popularity.

Developments in information technology have also had an impact on transport demand for work and social activities. For example, two decades ago, these IT practices weren’t common: email, video-conferencing services, data-sharing, e-commerce, online delivery services, social media platforms, and digital streaming services.

In the short-term, DPTI needs urgently to recalibrate and validate its existing models to current data. These models were last calibrated using data from the 1999 Metropolitan Adelaide Household Travel Survey (MAHTS). That data is now nearly twenty years old, and likely not reflective of current or future travel patterns within the region.

Recalibrating and validating existing (and new) models

Researchers detailed how the department could use current data to urgently recalibrate and validate its existing models. Here are the four steps:

1. Develop a business case and lobby the government to secure funding and appropriate services from the private sector to collect new data from households in the Greater Adelaide metropolis.
2. Harness GPS-based smartphone surveys to replace household travel diaries for data collection.
3. Avoid using standalone GPS loggers to complement or replace existing data collections. This technology will cost more than existing methods.
4. Collect field data from the new household travel survey in the second year of the four-year plan.

Additionally, the researchers suggested the department plan to regularly update its transport data models in the long term, so the same issues do not arise. The department could consider taking these actions:

• Replicate other studies that have shown cell phone CDR data as being able to reliably estimate origin-to-destination flows within a region
• Look for inspiration to household travel survey data from other jurisdictions, including capturing mode, route, and time-of-day choice
• Harness network-level data collected through Bluetooth sensors and public transport smart card automated fare collection systems. These data can recalibrate and validate the parameters of transport demand models, and
• Keep watch on how others are using emerging data sources, including CCTV footage and social media platforms.

Recommendations

The high-level recommendations emerge from the department adopting this framework:

• Identify a long-term vision
• Set measurable goals and targets
• Develop potential future scenarios
• Identify potential projects, and
• Assess and prioritise projects.

Check also the ‘findings’ section above as it offers insights into recommendations.

The report offered these specific recommendations about how the department could improve relevant organisational practices:

Model ownership and governance:

• Create a steering group to oversee the governance of the department’s modelling tools
• Outsource to the private sector model development but keep some in-house expertise, and
• Ensure potential project costs include funding for model improvements

Engagement between model developers and end users:

• Incentivise those developers to engage early and often with stakeholders
• Use the model to measure risk and uncertainty, too, and
• Develop consistent documentation explaining how to use the department’s suite of modelling tools.

Expected project impacts

One of the key recommendations from this study was that DIT’s strategic models need urgently to be recalibrated and validated using current data. However, the collection of household travel survey data was deemed too expensive. As an alternative, a follow-up piece of research was undertaken by DIT, UniSA and iMOVE examining the viability of ‘big data’ sources, such as mobility information from Bluetooth sensors and public transport smartcards, to replace traditional survey datasets for strategic modelling.

Findings from the study have been used in the recalibration of DIT’s strategic model and have been published by the journal Transportation Research Part A: Policy & Practice.

Associate Professor Akshay Vij – University of South Australia

Download the report

Download your copy of the wrap-up report, Rethinking strategic transport planning practices in South Australia, by clicking the button below.

DOWNLOAD THE REPORT