Forecasting the Impact of Active Travel Interventions
By:
TfSE - Joshua Jiao
Walking and cycling deliver substantial economic and environmental benefits, with far-reaching implications for public health and urban sustainability. According to the Sustrans Walking and Cycling Index 2023 report [1], active travel in UK cities prevents over 21,000 serious long-term health conditions annually, generating economic benefits worth £6.1 billion. Beyond economic value, walking and cycling contribute to environmental improvements by reducing air pollution. A 2022 study of Low Traffic Neighbourhoods (LTNs) in the London Borough of Islington, for instance, reported a 5.7% reduction in nitrogen dioxide (NO₂) levels at internal sites and an 8.9% reduction at boundary sites [2].
However, realising these benefits hinges on the ability to robustly forecast demand for walking and cycling. Robust demand forecasting is critical for policymakers and planners to assess the cost-effectiveness of interventions. A well-informed approach ensures that public funds are allocated to projects with the highest potential impact. Similar to forecasting highway traffic flows, forecasting active travel usage involves two key components: demand estimation and network modelling.
Forecasting Active Travel Demand
Department for Transport (DfT)'s Transport Analysis Guidance (TAG) Unit A5.1 Active Mode Appraisal [3] outlines three key methodologies for estimating active travel demand:
1. Comparative Study
A comparative approach evaluates proposed interventions against similar, existing schemes, deriving demand estimates based on observed outcomes. This method is straightforward and applicable to various interventions, however, identifying a truly comparable scheme can be challenging. Factors such as socio-economic differences and external influences must be carefully considered.
For instance, the London Millennium Footbridge was expected to attract approximately 10,000 daily users, whereas the York Millennium Bridge, despite increasing active travel usage by nearly 70%, records an average of only 2,000 daily users. These discrepancies highlight the importance of contextual factors in comparative analysis. [4]
2. Sketch Plan Method
The sketch plan approach utilises simple, rule-of-thumb calculations based on readily available data, such as census travel-to-work matrices, travel distances, and demographic forecasts. It provides approximate demand estimates using established elasticities from previous studies.
For example, TAG Unit A5.1 suggests an elasticity of +0.05 for changes in cycling demand relative to the proportion of road infrastructure dedicated to cycling. Suppose a district currently has 2,000 daily cycle trips, with 10% cycle-friendly infrastructure. Introducing a 10-kilometre off-road cycle route would increase the infrastructure proportion to 12%, resulting in a projected 1% increase in cycle trips—equivalent to 20 additional daily trips.
While this approach is relatively easy to implement, its focuses on infrastructure length limits its applications, particularly when distinguishing between different intervention types, such as segregated cycle lanes versus painted cycle markings.
3. Four-Stage Strategic Modelling
The most comprehensive forecasting method utilises strategic four-stage transport models, predicting changes in trip generation and mode share by analysing travel utility factors such as time, cost, and user preferences. This approach is detailed in the TAG Unit M2.1 Variable Demand Modelling [5], it requires detailed input data, making it resource-intensive but capable of delivering robust, scenario-based insights.
A notable example is Transport for London’s (TfL) MoTiON model [6], which forecasts travel behaviour changes in response to infrastructure improvements, policy shifts, and external factors such as car ownership levels. Built using data from the London Travel Demand Survey, MoTiON enables evidence-based decision-making at a strategic level.
4. Emerging Techniques: Microsimulation and Agent-Based Modelling
While not explicitly mentioned in the TAG Unit A5.1 Active Mode Appraisal, microsimulation and activity-based modelling approaches are gaining popularity. Models such as the activity-based multi-modal model developed by the Cambridgeshire and Peterborough Combined Authority, simulate trips based on demographic factors and trip characteristics, providing granular insights into active travel uptake. However, developing these models requires detailed travel diary data and substantial sample sizes to achieve robust accuracy.
Modelling Active Travel Networks
On the network side, a critical challenge in active travel modelling is the representation of travel costs. Both strategic four-stage models and microsimulation approaches rely on an accurate representation of travel times and costs using the concept of generalised costs, which incorporate both monetary and time-related costs into a single metric. Additionally, they allow different user groups (e.g. commuters vs. business travellers) to have different values of journey time.
However, unlike highway users, where congestion is a primary determinant, active travel decisions are influenced by factors such as surroundings, perceived safety, traffic volumes, and route gradient (hilliness). Failure to incorporate these elements into generalised costs may result in models that cannot differentiate, for example, between a traffic-free cycling highway and a standard cycle lane on a major road.
One of the potential solutions is known as Level of Traffic Stress (LTS). Researchers at the Mineta Transportation Institute (USA) have developed a four-point framework [7], classifying cycling routes into categories based on perceived stress levels:
LTS1 - comfortable for all ages and abilities
LTS2 - comfortable for most adults
LTS3 - comfortable for confident bicyclists
LTS4 - uncomfortable for most
Please click link here to view the photo reference: Level of Traffic Stress — What it Means for Building Better Bike Networks | by Alta | Alta
Scholars from the University of Cambridge then have introduced detour factors to represent route attractiveness based on empirical studies across 36 cities [8]. These factors allow the LTS to be interpreted in terms of longer perceived distances. For LTS1 (i.e. the safest route), the factored distance is equal to the actual route distance. For LTS2-4, additional factors are applied to the actual route distance to reflect an increased generalised cost to the cyclist.
User Segmentation in Active Travel Modelling
Also, Active Ttravel forecasting requires segmentation beyond traditional trip purpose-based categorisation. Differences in safety perceptions by gender, age, ethnicity, and car ownership significantly influence travel choices [9]. For example, female cyclists may prioritise safety more than their male counterparts, impacting uptake rates and route preferences.
Choosing the Right Approach
While no single methodology is universally applicable, the choice of approach should be proportionate to the scheme’s objectives and complexity. As summarised below, each approach has its strengths and challenges.
Approach: Comparative Study
Implementation Level: Easy
Best for: All interventions
Challenges: Finding a comparable case study can be challenging
Approach: Sketch Plan
Implementation Level: Medium
Best for: Increasing infrastructure length
Challenges: Relies on distance-based assumptions
Approach: Four-Stage Model
Implementation Level: Hard
Best for: Most interventions
Challenges: Requires detailed input data; constrained by how the active travel network and travel cost are modelled
Approach: Microsimulation/ABM
Implementation Level: Hard
Best for: All interventions
Challenges: Resource and data intensive; constrained by how the active travel network and travel cost are modelled
Case Studies to Complement the Clinic
We are seeking views on whether a Clinic to complement this blog would benefit you. The session will provide a practical, interactive space to turn ideas into practice and tackle real-world challenges using your case studies as live examples.
If you would like to participate, see value in holding the session, or have a case study to share, please respond by 14 February 2025 by emailing coe@transportforthesoutheast.org.uk.
Let’s work together to improve modelling approaches and share insights that support better active travel outcomes.
References
[1] Sustrans, “The Walking and Cycling Index,” 2023.
[2] IPSOS, “Low Traffic Neighbourhoods,” 2024.
[3] Department for Transport, “TAG unit A5-1 active mode appraisal,” 2022. [Online]. Available: https://www.gov.uk/government/publications/tag-unit-a5-1-active-mode-appraisal.
[4] Department for Transport, “Encouraging walking and cycling: Success Stories,” 2004.
[5] Department for Transport, “TAG unit M2-1 variable demand modelling,” 2024. [Online]. Available: https://www.gov.uk/government/publications/tag-unit-m2-1-variable-demand-modelling.
[6] Strategic transport models - Transport for London
[7] M. C. Mekuria, P. G. Furth and H. Nixon, “Low-Stress Bicycling and Network Connectivity,” Mineta Transportation Institute Publications, 2012.
[8] R. Cervero, S. Denman and Y. Jin, “Network design, built and natural environments, and bicycle commuting: Evidence from British cities and towns,” Transport Policy, vol. 74, pp. 153-164, 2019.
[9] J. Brainard, R. Cooke, K. Lane and C. Salter, “Age, sex and other correlates with active travel walking and cycling in England: Analysis of responses to the Active Lives Survey 2016/17,” Preventive Medicine, vol. 123, pp. 225-231, 2019.
About the Author
Dr Joshua Jiao is an award-winning transport modeller with over 10 years’ experience in transport planning and modelling. He joined Transport for the South East as Analysis Manager in late 2023, playing a key role in developing analytical frameworks and transport models for regional and national strategies. Previously, he worked in private sector on models’ development and scheme appraisal projects, including active travel models for the Liverpool City Region, strategic models for Thurrock, and the Lower Thames Crossing Area Model and outline business case. Recognised for his research, Joshua received the Neil Mansfield Memorial Award at ETC 2019 for his PhD at the University of Cambridge on travel demand modelling. Passionate about data and innovation in transport planning, he has also led projects such as Artificial Intelligence for Mass Model Automation and Review of Plug-in-Vehicle Uptake in Scotland.