Active transport in regional areas: Bonus expert Q&A

Last week’s Encouraging active transport in regional areas webinar was so well attended, and attracted so much participation in the Q&A section, that some questions asked by attendees went unanswered. Here the three webinar guests answer those questions.

We’d again like to thank those that attended for their interest and their questions, and our three webinar guests:

• Bruno Spandonide: Department of Transport and Planning, Associate Director, Transport Strategy (Hume, Lodden and Mallee)
• Matthew Kerlin: Coordinator: Active Communities, City of Greater Bendigo
• Dr Afshin Jafari: Research Fellow, Centre for Urban Research, RMIT University

Please find below the bonus Q&A material. If you’d like to watch the webinar recording it’s available at Webinar: Encouraging active transport in regional areas.

What differences in AT infrastructure needs do you see between regional and outer suburbs areas? (Wendy Nash)

One key difference between regional cities and the outer suburbs of capital cities is that, in regional areas, the distances for most work and non-work trips fall within a feasible cycling range. As a result, there is significant potential for increased cycling uptake. In terms of infrastructure. Bendigo has a growing number of high-quality cycling facilities in the inner parts of the city, however, the main challenge remains connectivity. (Afshin)

Do you have open/available data defining those cycle networks? (Tim Doze)

The city’s protected cycling paths are one of the layers as part of the mapping data at City of Greater Bendigo : Cycling – Everyday Rides (Matthew).

How much have you focused on communications with the local community engagement? (Wendy Nash)

For this project, it has been a survey mapping preferred rides and challenging locations to inform the modelling. For the city’s Protected Walking and Cycling Network we work closely with our local advocacy group (Bike Bendigo) and have used a multimedia comms approach, a mix of hard copy magazines, social media, and mailouts for individual projects. (Matthew)

Great to see the statistics about women in cycling, however I am curious if there was any parameters around driveways and conflict points with regards to the 87% of wome who are comfortable riding on a shared pathway? I know this is a major issue/consideration when considering seperated cycling infrastructure in the road verge. (Dylan Bedward)

Your question was phrased around where you feel comfortable riding. As the majority of the city’s current shared paths are off-road (along creekways, etc.) I think that this would likely have influenced the response, with the key conflict points being the road crossings along these paths (we heard that road crossings repeatedly in the engagement as a barrier to walking and cycling).

Our Mundy Street shared path will cross driveways and part of our evaluation will be asking about our design solution and treatment, so we’re hoping to get more info on this as we aim for continuous improvement as we roll out more shared paths across the city. (Matthew)

Just wanting to check that car travel time would not be included as a benefit in this model. (anonymous)

No, in the current version of the model, we are focused on changes to cycling level of traffic stress resulting from new cycling infrastructure and any uptake that follows. (Afshin)

ABMs are traditionally data hungry and can take a significant time to run a single scenario, not to mention then analysing/digesting the results. You mentioned you’re keen for this to be a tool that can change the way we plan and iterate more effectively. Have you been able to (or planning to) overcome the challenges of run times to ensure it doesn’t just become another blackbox model that takes significant amount of time to run? (anonymous)

This is a very valid point and something we are spending a significant amount of time on both improving the efficiency of the model and designing a modular, scenario-testing ABM for the backend of our online tool. In this setup, the baseline won’t be rerun each time a user makes a change and instead, we will only recalculate what is most relevant to the user’s changes and the selected outputs for analysis. More details will be shared next year in our second webinar on this project. (Afshin)

Is LTS a mid-block assessment only? How does LTS take into account junction delays at intersections across the network? (anonymous)

LTS was calculated for both road segments and intersections. Intersection LTS was only considered during cycling route assignment, road segment LTS was used in both the routing and the mode choice model. (Afshin)

Does the demand generation take into account ‘work from home’ for full-time workers? (anonymous)

The process extracts activity patterns from the travel survey data. We do not explicitly model working from home, but since the travel survey includes full-time workers with no recorded work trips on the survey day, our activity-based demand also includes full-time workers without a work trip, which are not common as the VISTA data used is from the pre-COVID period. (Afshin)

For the demand generation comparison, did you use weighted or unweighted VISTA?  (anonymous)

We used weighted VISTA. (Afshin)

How was travel time calculated? Average speed, or different speeds on different routes at different times of the day? (anonymous)

For cars, we currently use 85% of the speed limit. In the next version, we will switch to a temporal network with time-dependent speeds to better capture peak-hour delays. For public transport, we use actual timetables from GTFS. Walking routes are based on the shortest path, assuming an average walking speed. For cycling, we use impedance-based routing with an average cycling speed. (Afshin)

would the sample size for the cycling map be considered statistically significant to be reliable? (anonymous)

Yes, the current sample size is expected to be sufficient for a route choice model with a small number of variables. (Afshin)

What are the preliminary insights from the Bendigo ABM and what have u shared with ur govt partners? how do the insights help them strengthen their arguments for funding? (Terry Graham)

The Bendigo ABM will serve as the baseline for cycling investment scenario testing. In the next phase of the project, we will implement functionality for partners to interact with the model through an online user interface, allowing them to test their own scenarios. (Afshin)

The model looks very good, and I understand this is about regional Victoria – but the Strategic Cycling Corridor has languished since 2020. Is this tool going to be used to change the best locations for it? Has it been abandoned? (Irene Moser)

The tool will enable testing of the expected benefits of the corridors, as well as identifying network gaps that may emerge from the implementation of SCCs. As such, it will serve as a complementary decision-support tool for these interventions. (Afshin)

Are you using street centre lines for modelling, or actual cycling path locations…  and how they work into different interesection types (roundabouts, etc.)? (M White)

We use line geometries from OSM to create the street network, applying some level of simplification, for example, consolidating intersections and roundabouts into single nodes, while preserving key attributes such as actual length and bike lane types. As a result, the network geometries closely resemble those found in commonly used online maps while ensuring run-time efficiency. (Afshin)

What will outputs of the model include? (Stephanie Miller)

For this project, we include mode shift and changes in exposure to different cycling levels of traffic stress as primary outputs. Building on these, our online tool will also provide estimates of environmental and health benefits, along with other commonly reported benefits, as recommended by active transport benefit estimation guidelines. More details will be shared in the webinar 2 (after the completion of the project). (Afshin)

What kinds of infrastructure treatments can be tested?(anonymous)

At this stage, the tool is being designed to support two types of interventions: adding bike lanes to existing road segments without bike infrastructure, and creating new off-road bike paths. More details will be provided in this project’s next webinar. (Afshin)

A powerful body of work Afshin, good work. How did you determine where to assign trips that aren’t utility based, for example joy rides? (anonymous)

Thank you. Joy rides or recreational trips are not included in this work, as they require a very different choice modelling approach for both route and mode choice. For someone riding for recreation, travel time is unlikely to be a key factor and instead attributes such as greenery or quietness of the path are likely to be more important. We did consider using data from Strava to capture such trips, but for the current tool and model the focus remains on cycling trips for transport purposes. (Afshin)

Have you considered the activities of commercial bikes in your model? E.g. cargo bike for transporting parcels like those operated by Australia Post. (anonymous)

We did not account for different types of bikes or those used during work activities, such as bikes used by Australia Post. The current model only includes bike trips for transport purposes, for example, travelling to work or to a shop. (Afshin)

What would be required to apply this methodoloy at the metro Melbourne scale, and is this part of the plan? (Jody Betzien)

This work builds on previous work undertaken for Greater Melbourne but has significantly advanced beyond it. Beyond applying these changes to Greater Melbourne which requires time and resources, it also presents a challenge in terms of the online tool’s runtime efficiency. We are addressing this by developing a modular design that will allow us to manage runtime even at larger scales, such as Greater Melbourne or all of Victoria.

That said, the main requirement is securing future funding to extend the work to other regions, as this project is currently focused on Greater Bendigo. (Afshin)

How scalable is this to other locations outside of Bendigo? e.g. in larger and more complex urban areas, such as Melbourne. (Sara Stace)

Excellent question! High scalability potential for cycling in terms of methodology requires a tailored agent-based model calibration model, requiring a lot more work in terms of walking integration and intermodal feedback loop opportunities. (Bruno)

How does your ABM compare with other similar models such as AIMSUN used in similar studies ? (M White)

Our modelling framework is activity-based and agent-based, built on the MATSim platform, making it more comparable to models like MABM or VITM2. As such, it is better suited to city-scale modelling and interventions. In contrast, micro-scale models such as those developed using AIMSUN are more appropriate for detailed interventions at the intersection level. (Afshin)

Only children 12 and under can cycle on the footpath, 13 and over years must cycle on the road. (Rachel Carlisle)

That is correct. However, the challenge is that the sample size for students in the VISTA dataset is very small. If we were to split students into those older and younger than 13, the number of cycling trips would be too low for the choice model to produce statistically significant results. Therefore, we combined all student cyclists into a single cohort, acknowledging that this approach affects the modelling of cycling infrastructure impacts, given that some students in this group are permitted to cycle on footpaths. (Afshin)