A Co-Created Framework to Define Digital Twinning Use Cases for Urban Transport Decarbonisation

Abstract

With global urbanisation anticipated to reach 68% by 2050, there is a significant risk of exacerbating urban transport emissions. Urban transport decarbonisation is a complex adaptive system challenge, the understanding and optimisation of which could be supported by digital twins (DTs). Although prior research has explored digital and big data technology applications, creating actionable insights requires human-centred designs. We conducted a structured workshop to gather practitioner views on how urban-scale DTs can support transport decarbonisation. Specifically, we explored the outcomes they aim to achieve, the interventions they are interested in, and the value digital twinning offers compared to current methods. The data was synthesised and analysed to identify (1) impacts, (2) interventions, (3) location types, (4) data sources and (5) feedback mechanisms of importance to participants. These five aspects are proposed as a framework to support the definition of digital twinning use cases targeting urban transport decarbonisation. Application of the framework encourages creators to explicitly consider the services to be provided to users, how the derived insights influence the real world and the data connections between the physical and digital, noting that these are often overlooked in reported research. A framework application is illustrated through an example use case described for the West Midlands, UK.

1. Introduction

1.1. Urban Transport Decarbonisation

Interventions to decarbonise urban transport are critical given: the necessity to limit human-caused global warming by reaching net zero emissions [1]; the high contribution of the transport sector to global emissions and its continued reliance on fossil fuels [2]; and continued urbanisation, which is projected to result in 68% of the global population residing in urban areas by 2050 [3]. To achieve sustainable urban mobility, urban planners and policymakers must consider the complex interdependencies between urban form, transport and energy systems when implementing interventions [4]. Alongside the electrification of vehicles, local policymakers are concerned with incentivising a modal shift from private cars to public and active travel modes in a fair, inclusive and equitable way to facilitate a just transition to decarbonised transport [5]. Reducing car dependency and its associated harms—including bodily physical harm, ill health, environmental damage and social harms [6]—is viewed as beneficial not only for the environment but also for individuals and wider society [7].

The range of potential interventions to decarbonise urban transport is vast. When considering implementation, policymakers are concerned with not only the technical and operational feasibility of individual interventions, but also how they may interact with each other and the resultant impacts on transport emissions and other factors. The Sustainable Accessibility and Mobility (SAM) framework [8] takes a hierarchical and human-centred approach to reducing transport emissions, requiring travellers to first substitute trips (avoid), shift modes if travel is necessary (shift) and switch fuels (improve). Planners can prioritise interventions that aim to encourage this hierarchy, for example, land use changes to create 15 min neighbourhoods, improving the availability and attractiveness of public transport (PT) to reduce car use, and implementing technologies such as EV-charging infrastructure to reduce trip carbon intensity (see Figure 1). However, understanding human decision making and changing travel behaviour is inherently complex, with decisions driven by an individual’s capabilities, opportunities and motivations [9].

1.2. Smart Cities

Digital and big data analytics technologies—including but not limited to sensors, wireless networks, cloud computing, machine learning (ML) and artificial intelligence (AI) [10]—are increasingly being used by urban transport planners to collect and analyse data on urban operations, citizen behaviours and the emergent characteristics of both under the smart city paradigm. This in turn has driven advances in urban operational and planning decision support. Standards such as the ISO/IEC 5087 series [11], which focus on foundational ontologies for city data, have evolved to support smart city decision making. However, smart city standards have not been able to support full decarbonisation of urban transport due to their inherent focus on technological changes over system changes. Many initiatives prioritise technological solutions such as electric vehicles (EVs), autonomous vehicles (AVs) [12] and smart traffic management [13], but these could increase private car use if not combined with policies that manage demand and prioritise PT use. Urban transport decarbonisation is a complex adaptive system challenge, wherein the ability to predict the emergent system response under a range of future scenarios, including policy implementation, is very complicated.

1.3. Urban Digital Twinning

Real-world urban systems, including transport, are increasingly being represented digitally, using models, shadows or twins. In this work, we define digital representations (DRs) based on Nag et al. [14], who combine model, shadow and twin definitions with the five-component definition proposed by Tao et al. [15]. The classification of a DR as model, shadow or twin is based on the level of data flow automation between the physical object, digital object (including data model) and the service components (see Figure 2a and

Table 1. General description of data flow automation between physical and digital objects for DRs—digital models, shadows and twins—and between the specific components shown in Figure 2a.

Data Flow	Model	Shadow	Twin
General description	No automatic data flow between physical and digital	Automatic data flow from physical to digital only	Automatic, bidirectional data flow between physical and digital
Physical object to digital object	Manual	Automatic	Automatic
Digital object to service component	Manual	Manual/Automatic	Automatic
Service component to physical object	Manual	Manual	Automatic

). Service refers to functional services for users, such as interaction, visualisation and metrics. According to this definition, digital twins (DTs) have automatic, bidirectional data flow between all components, including between the service component and the physical object, enabling automatic real-world control. Such control is not possible with models or shadows. This broadly aligns with the definitions provided by the UK National Digital Twin Programme, NDTP (https://www.ndtp.co.uk/ (accessed on 21 November 2025)), with automation of bidirectional data flow implied. The NDTP states that transport DTs require complex multi-fidelity twins with (i) real-time inputs to accurately represent the environment and (ii) the ability to output to infrastructure in that environment.

Urban DTs are proposed to leverage technologies to facilitate bidirectional data flows between urban physical systems and their digital counterparts, enabling monitoring, interaction and control to improve day-to-day operations [16]. They are also envisaged to facilitate scenario modelling, providing a safe environment in which to test the likely real-world, emergent impacts of proposed interventions to urban structure and policy, generating evidence to support longer-term decision making. This extends the smart city concept of collecting and analysing data for operational and planning decision support. With the capability to determine and influence urban dynamics at both short-term and long-term scales, and integrate diverse data sources, urban DTs are expected to provide a more holistic understanding of travel behaviour to support transportation planning and policymaking [17]. User interaction is critical, requiring that urban-scale DTs enable multiple users to interact with and interpret data (at different temporal scales) to create value for a range of functional uses [18]. For urban transport planning, the main users are expected to sit within government agencies, but may extend to non-governmental stakeholders including infrastructure managers, operators and regulators.

Nag et al. [14] conclude that many so-called DTs for transport planning are models or shadows lacking automatic or bidirectional interactions. This is unsurprising given that, for longer-term scenario modelling, automatic feedback to the real world is not necessarily feasible or desirable, and digital models or shadows (that provide insights to inform human decision making) may be more appropriate. We introduce the concept of “digital twinning” to reflect this, whereby urban systems are digitally represented using an ecosystem of interrelated DRs (models, shadows and twins) that together deliver operational control, decision support and scenario modelling functionalities, ideally in a single platform.

1.4. Context of This Research

We set out to determine with practitioners what specific functional uses are of interest for an urban-scale DT for transport decarbonisation. This was to support the definition of use cases for the Digital Twinning Research Hub for Decarbonising Transport (TransiT) (https://transit.ac.uk/ (accessed on 21 November 2025)) prototype multimodal passenger transport DR of the West Midlands, UK, which is currently being developed. However, the results are expected to be of broader interest to the research community.

Specifically, we sought to answer the following research questions:

• From the perspective of practitioners, which impacts, interventions and location types are of priority concern for transport decarbonisation?
• Which data sources and feedback mechanisms do practitioners see as the “value-adding” contributions of digital twins?
• How can these aspects be combined into a framework for defining digital twinning use cases?

We held a structured workshop with experts from local authorities and industry working on transport policy (in the West Midlands and wider UK context) to capture their perspectives on the desired outcomes from urban transport decarbonisation interventions, which interventions are of interest and where they should be applied, and the added value they expect from digital twinning over existing methods. The data has been transcribed, synthesised and analysed to create a framework comprising five aspects that can be applied to define practical digital twinning use cases targeting urban transport decarbonisation.

A bottom-up approach is needed, given that few studies consider the services provided to users from a practitioner’s perspective (with most either proposing what the created technology could be used for or not considering user services at all) [14]. Few studies explicitly document stakeholder expectations from transport system digital twinning. Callcut et al. [19] combine interviews, surveys and a literature review to understand digital twin applications in civil infrastructure systems, including but not limited to transport. They highlight a need for digital twin implementation to be more purpose-driven, delivered alongside those who have knowledge of relevant issues. Belfadel et al. [20] conducted workshops to identify the expected outcomes and impacts of digital twins supporting the creation of a three-layer digital twin framework, but this specifically targeted city logistics, not broader urban transport. Even in collaborative research projects creating city-scale twins, there is often minimal detail available regarding how use cases were defined. For example, METACITIES will build a twin of Patras (Greece) to test smart mobility policies and solutions related to parking, the environmental impact of traffic incidents and emergency management, but does not state how targets were agreed on [10]. Digital Urban European Twins (DUET) created digital twin pilots for Athens (Greece), Flanders (Belgium) and Pilsen (Czechia) to explore urban planning scenarios related to mobility, the environment and health based on gathered user stories [21]. Although the scenario specifications are reported, descriptions of how they were agreed are cursory and no underlying data is given. We hope by sharing our methods and resultant framework, we can support others in collaboratively defining practical applications of digital twinning to support urban transport decarbonisation, focussing on real user needs.

1.5. Overview of Paper Structure and Contributions

The paper is organised as follows: Section 2 describes the methodology, including how the workshop was structured and the data analysed to develop the framework; Section 3 presents results from the different workshop exercises relating to the digital twinning context, outcomes, use cases and added value of digital twinning, as well as summarising the framework described above; Section 4 discusses the results and Section 5 concludes the paper.

The paper contributes the following:

• Summarises from a practitioner perspective the impacts, interventions and location types of priority concern for transport decarbonisation
• Provides context regarding the limitations of existing transport models that digital twinning should aim to overcome.
• Captures the data sources and feedback mechanisms that practitioners see as the value-adding contributions of digital twins.
• Brings together the five aspects (highlighted in bold above) as a framework to support the definition of digital twinning use cases for urban transport decarbonisation.
• Illustrates application of the framework through an example related to urban transport decarbonisation in the West Midlands, UK.

2. Methodology

This section describes the techniques and procedures used to answer the research questions. Practitioner perspectives on digital twinning use cases for multimodal transport decarbonisation in the West Midlands were gathered through a structured workshop held in person at the University of Birmingham in 2025. The workshop was attended by 18 individuals working across academia and industry in diverse roles, as well as local authorities, invited due to their transport-related expertise in policy, decarbonisation and modelling (see Appendix A for further information). The overarching research method may be considered Design Science Research (DSR), which seeks to generate knowledge related to the design of artefacts to solve real-world problems [22], in this case, digital representations to support urban transport decarbonisation. In relation to the DSR process model [23], the work is focussed on Activity 2—define the objectives for a solution, i.e., understanding what a better transport digital representation would accomplish for practitioners. This work contributes to the design of the prototype DR. Once the prototype has been developed, its performance will be empirically evaluated.

2.1. Framework Development

The workshop data was transcribed, synthesised and analysed, using reflexive thematic analysis [24] where sufficient responses were received, to determine (1) impacts, (2) interventions, (3) location types, (4) data sources and (5) feedback mechanisms of importance to participants. Together these aspects form a framework with which to define practical use cases that apply digital twinning to address real-world transport issues.

These aspects were explicitly targeted to overcome gaps identified in the literature with respect to understanding the services desired by users and the automatic data flow that differentiates a DT from a model or shadow. The impacts and location types provide an understanding of the outcomes desired from local transport planning and projects, e.g., reduced air pollution in the city centre, whilst the interventions suggest solutions that may achieve these outcomes, e.g., low-emission zones. Data sources and feedback mechanisms capture the opportunities for digital twinning, indicating the types of real-world data that are typically lacking in existing models, e.g., PT occupancy, and where automated, real-time control or feedback, e.g., via variable message signs and speed limits on smart motorways, could support the realisation of the desired outcomes. It should be recognised that the desired data may not exist in current models because there are no reliable, real-time sources that capture it.

Figure 2 maps the different aspects to be captured during the workshop discussions to a generalised digital representation (based on Nag et al. [14]) where the data flow between components can be either manual or automatic:

• Impacts, interventions and location types correspond to the service component, which provides a means for user interaction. In scenario modelling, users specify the interventions and locations to be modelled and assess the impacts with respect to a range of metrics.
• Data sources correspond to the link from the physical object to the data model and digital representation. As per the NDTP, real-time inputs help accurately represent the environment.
• Feedback mechanisms correspond to the link between the service component and physical object, indicating how the insights derived from the digital representation can impact the real-world state of the transport system, which in turn will update the digital representation. As per the NDTP, this creates the ability to output to infrastructure in that environment.

2.2. Workshop Purpose

The workshop was designed to support TransiT in identifying use cases to prioritise for exploration using a prototype DR of multimodal passenger transport in the West Midlands, UK. The prototype is expected to be developed and refined within the funded period (ending June 2029). The Carbon Assessment Playbook (CAP, https://qcrinfo.wordpress.com/ (accessed on 21 November 2025)), developed by England’s Subnational Transport Bodies, sets out 29 local transport planning interventions categorised as avoid, shift or improve, which relate to behavioural change, integrated planning policy, active travel, parking charging and traffic management, public transport, technology, and low emission vehicles. Intervention cards summarise the potential carbon reduction and co-benefits of each intervention and users can assess the impact of applying interventions to a specific area using the online Policy Builder. Published reports suggest interventions that may have the biggest impact on emissions for local authorities in the West Midlands and other regions of England. However, many of these interventions (particularly those related to modal shift) have limited evidence, potential secondary emissions and underestimated impacts. With such a wide range of possible interventions and application locations (the West Midlands region comprises 14 local authorities) the team sought input from practitioners with local expertise to determine the most beneficial targets for the prototype development in terms of added value over existing methods.

Five digital twinning use cases for the prototype West Midlands transport DR were defined by applying the framework developed from the workshop outputs, augmented by researcher knowledge. The use cases were shared with workshop attendees who ranked them to gain consensus on the use case to pursue for prototype demonstration. We present one of the use cases in Section 3.6 to illustrate the application of the framework. We believe the insights derived from the workshop are of broader interest to the research community and can help others also define practical applications of digital twinning to support urban transport decarbonisation that focus on real user needs.

2.3. Workshop Procedure and Data Collection

Attendees were first introduced to TransiT and the objectives of the proposed prototype West Midlands DR and the CAP interventions, both generally and as reported for the West Midlands, to contextualise the workshop. The presentation covered the workshop purpose, data collection methods and how the data would subsequently be used to inform research and outputs, including academic publications. All participants signed a consent form confirming voluntary participation and agreeing to anonymised data collection for the purposes described. Attendees completed individual warm-up exercises on a paper worksheet (without providing any identifying information) and participated in three group exercises with responses captured on Post-it notes and flip chart paper. Once transcribed, it was not possible to trace any of the responses back to individual attendees. The workshop was conducted, and all data handled, in accordance with the ethics review submitted to and approved by the Durham University Engineering Ethics Committee. All responses are available in the Supplementary Materials, with locations specific to the West Midlands generalised to ensure broader application, e.g., commuter railway lines or arterial roads indicated by italics.

The warm-up exercises comprised the three questions below to capture initial perspectives from attendees and contextualise the workshop.

• How do you identify the problems to address today?
• What added value do you want to see over existing methods?
• What interventions are you most interested in?

Exercise 1 built directly upon the third warm-up question, asking attendees to generate outcomes and use cases. Outcomes were intended to capture the impacts practitioners wanted to see from local transport interventions, in specific locations if applicable. Attendees were given example outcomes, including “reduce carbon emissions in the city centre” and “reduce congestion on X arterial road during peak hours”. Use cases were intended to capture the interventions that might achieve these outcomes. Attendees were given examples such as “make public transport free” and “operate a 24-h tram service between X and Y”.

Exercise 2 built upon the second warm-up question by asking attendees to think specifically about the added value of digital twinning through the following questions:

• What dynamic/time-dependent data can we add to our models?
• Where could we add closed feedback between the physical and digital assets?
• Are there any interventions for which we could compare traditional and digital twinning (TransiT) methods?

Exercise 3 provided an opportunity for attendees to document additional outcomes and use cases prompted by Exercise 2.

2.4. Data Analysis

The responses generated by Exercises 1 and 3 were analysed together, as only a small number were added during Exercise 3. A few outcomes were reclassified as use cases and vice versa based on whether they described impacts or interventions. Outcomes and use cases were analysed separately. The transcribed responses were imported into Miro (https://miro.com (accessed 21 November 2025)), an online visual workspace, for researchers working on the TransiT West Midlands transport DR to undertake collaborative affinity diagramming (where similar concepts are grouped together to identify themes [25]). This activity supported the initial steps of systematic thematic analysis as set out by Naeem et al. [26], which requires familiarisation with the data. Since the written responses were generally brief, we either used the whole response as a quote, or split it into multiple quotes when there was an obvious difference in focus. For example, many attendees used bullet points to list multiple outcomes or use cases on a single Post-it; in these cases, each bullet point was treated as a separate quote. Over two sessions totalling 3 h, 6 TransiT researchers grouped quotes together and named the groups. The team included two members who were not present at the workshop. A single researcher, who was present, then continued reflexive thematic analysis. Keywords were assigned to each relevant quote, based on repeated words (step 2), and then codes were developed (step 3) inductively based on emergent patterns in the data. Codes were then grouped together to create themes (step 4) that relate back to the research questions, again inductively. The keywords, codes and themes reflected the ideas developed collaboratively through affinity diagramming and were shared back with the team and accepted without disagreement or modification. We did not go on to develop conceptual models (as described in steps 5 and 6), instead using the two sets of themes and codes generated, alongside synthesised data from the other exercises, to develop the framework.

Responses to each of the warm-up questions and exercise 2 questions were analysed separately, with quotes selected and keywords assigned as described above to synthesise the main findings. Codes and themes were not developed due to the relatively limited number of responses.

The data was summarised as five aspects—impacts, interventions, location types, data sources and feedback mechanisms—that form a framework to support the definition of practical digital twinning use cases for urban transport decarbonisation.

3. Results

Figure 3 indicates how the data collected from the different exercises during the workshop were synthesised and analysed to derive the different framework aspects, and where in this Section the results are presented. An example is provided to illustrate an application of the framework to inform the development of use cases in Section 3.6.

3.1. Context

The warm-up exercises collected individual responses to three questions to focus participants on the subsequent group exercises. Together these questions help to contextualise the expectations of digital twinning for urban transport decarbonisation. Anonymised responses can be found in sheets 1–3 of the Supplementary Materials Excel spreadsheet. The assigned keywords correspond to the bullet point headings in each subsection below.

3.1.1. How Transport Issues Are Currently Identified

Transport issues are currently identified through policy, qualitative feedback from passengers and stakeholders, review of academic, grey literature and existing projects, and quantitative analysis of a wide range of data sources. This information informs the outcomes desired from local transport planning and projects and can help guide solutions, e.g., passengers may identify a lack of connectivity between two areas of the city and express a preference for bus services.

• Policy: Transport priorities can come top-down from policy at the government level, e.g., transport decarbonisation [27], and the mayoral level, e.g., journeys for everyone [28], or may be influenced by policy in other areas, e.g., new housing or employment creates changes in transport demand.
• Passenger feedback: Qualitative and quantitative feedback from transport user groups is important to understand the services users want and likely behavioural change. This can be captured through focus groups, surveys, and real-time walk-throughs of a journey.
• Stakeholder feedback: Other transport stakeholders can help elucidate problems, including infrastructure providers, operators, drivers, and maintainers. Stakeholders, including politicians, can provide input on the feasibility of proposed solutions, including but not limited to operational, financial and political feasibility.
• Literature and project review: Published research from academia and industry can provide insight into problems. This includes parallel projects in other sectors, e.g., freight and model or tool development internally and externally.
• Analysis of real-world data: Quantitative insights can be driven by real-world data on the current and historic state of transport, such as journey times, delays, reliability, incidents and failures, passenger numbers, mode split, car usage, emissions, safety, etc. Understanding where emissions originate in terms of trip distance, purpose and mode is important to identify areas with the greatest potential to shift or reduce car use. Demographic and economic data can also be used to infer changes in travel demand and behaviours. Data may be taken from open feeds or from an organisation’s or stakeholder’s own records or sensors. Data from trials, e.g., of pricing changes, is useful for determining the effect of interventions and whether this matches models.

3.1.2. Added Value Desired over Existing Methods

With respect to added value over existing methods, attendees want the ability to conduct both longer-term scenario modelling and generate real-time insights, which aligns well with the intended functionality of digital twins. They highlighted the need for modelling to be multimodal, capturing the wider effects of interventions on the whole transport system, and for it to incorporate realistic passenger behaviour. Whilst current models deliver this functionality to some extent, it is anticipated that digital twins will be capable of providing greater insights. Responses related to tool quality and data reinforce the importance of user interaction and consideration of how developed tools can practically support their users.

• Ability to conduct multimodal what-if analysis: Attendees want to be able to optioneer and test an intervention or multiple interventions, before real-world application, and be able to gauge the wider impacts across transport modes. This includes interactions between passenger and freight transportation, for example, recognising that the pseudo-randomness of freight demand is difficult to represent in conventional models. Analysis of spatial and place-based differences, e.g., urban vs. rural, is essential.
• Ability to generate real-time insights: Attendees want to be able to draw insights from near real-time data (e.g., delays, reliability, capacity, occupancy) to predict demand forecasts and inform operational and contingency planning. However, due to the complexities of incorporating real-time data, careful consideration should be given to understanding where it can truly add value.
• Inclusion of realistic passenger behaviour: Attendees want the interventions tested to have real-world applicability by considering passengers’ needs and behaviours, e.g., trip chaining within and between modes, perceived barriers to using PT and active travel. It is important to validate whether perspectives match reality and the potential impact this has on solution acceptance.
• Improved tool quality and robust data: Attendees want a tool that is user-friendly with easy-to-understand reports, and for intervention scenarios to be quick to produce and run. Reporting of potential impacts should reflect real-world key performance indicators. Any tools should make existing data, including historical data, easier to understand and use, e.g., provide higher resolution and demonstrate robustness in data about mobility and emissions. Data that is not currently being considered, such as asset information and supply side issues (e.g., for the railway—unavailability of rolling stock, staff, spare parts for maintenance), could be incorporated.

3.1.3. Transport Decarbonisation Interventions of Interest

This question provides insight into solutions that are at the forefront of attendees’ minds. They cover shift measures related to disincentivising car use and increasing the availability and attractiveness of PT and electrification as an improve measure to reduce modal emissions. It should be noted that switching to private electric vehicles does nothing to alleviate congestion, which can reduce the attractiveness of modes such as buses.

• Pricing: To both incentivise PT use and penalise car use: free or low-cost PT, integrated ticketing or ticket pricing reform, congestion charges, parking and pricing controls, and the introduction of individual carbon credits.
• Road space reallocation: To make space for active travel routes or create low-traffic neighbourhoods. For the latter, understanding disruption to traffic and whether this increases the overall car kilometres travelled.
• Meeting demand with public transport: Reopening local rail stations or opening new lines, running new routes, improving connectivity between PT modes, increasing service frequencies, offering services that reflect how people travel, e.g., trip chaining, meeting “accessibility desires”, e.g., no more than 10 minutes’ walk away from a PT stop or station.
• Public transport user experience: Improving PT reliability and information provision to customers, considering realistic operational simulation and what happens under disruption, e.g., short-term roadworks, rail closures, and meeting accessibility needs (step-free access, space for wheelchairs, pushchairs, and luggage). Considering not just passengers but also other people in the transport system, such as drivers and operators.
• Electrification: Electrification of public and private vehicles and the impact of electric vehicle-charging requirements on the grid in areas with high transport use.

3.2. Outcomes

Around 40 outcomes were captured across Exercises 1 and 3, defined as impact at location. Only eight responses gave specific locations, despite more than half the attendees being local to the West Midlands, so it is assumed that the outcomes are desired more broadly across the region. Table 2 illustrates how keywords were extracted from example quotations and linked to codes and themes. Figure 4 visualises the proportion of responses assigned to each theme, code and keyword. All anonymised responses can be found in sheet 4—Outcomes (1&3)—of the Supplementary Materials Excel spreadsheet, with keywords, codes and themes assigned. The results suggest attendees are interested in:

• The reduction in regional transport carbon emissions from a system perspective, primarily through modal shift but also through the reduction in emissions from individual modes.
• Increased sustainable transport use through better transport provision and improved transport user experience.

Table 2. Thematic approach used for outcomes generated during the workshop; brackets indicate the number of responses assigned to each theme, code and keyword.

Themes	Codes	Keywords	Example Quotations
1. Reduce transport emissions (10)	Modal shift (8)	Reduce car use (2)	“No cars in low emission zone”
		Increase use of Public Transport (PT) and active modes (6)	“Use public transport to travel to medical appointments”, “Make public transport more attractive to shift car journeys”
	Reduce mode emissions (2)	Reduce mode emissions (2)	“Reduce emissions of buses”, “Electrification of rail”
2. Encourage sustainable transport use (31)	Transport user experience (13)	Accessibility (1)	“Make public transport more accessible for wheelchair users, disabled, vulnerable groups”
		Capacity (4)	“Reduce occupancy of buses during peak times”, “Reduce station congestion from school drop off”
		Reliability (6)	“Improve customer confidence in public transport”
		Ticketing (2)	“Improve ticketing payment system”, “Reduce fare”
	Transport provision (18)	Match demand (5)	“Understand the ability of the system to absorb more users (long term and event based)”, “Understand demand movements”
		Alternatives to private cars (6)	“Increase public transport provision from rural areas”, “Reduce transport deserts”
		Connectivity (7)	“Improve transport connectivity”, “Social hubs interconnectivity improvements”

Table 2. Thematic approach used for outcomes generated during the workshop; brackets indicate the number of responses assigned to each theme, code and keyword.

Themes	Codes	Keywords	Example Quotations
1. Reduce transport emissions (10)	Modal shift (8)	Reduce car use (2)	“No cars in low emission zone”
		Increase use of Public Transport (PT) and active modes (6)	“Use public transport to travel to medical appointments”, “Make public transport more attractive to shift car journeys”
	Reduce mode emissions (2)	Reduce mode emissions (2)	“Reduce emissions of buses”, “Electrification of rail”
2. Encourage sustainable transport use (31)	Transport user experience (13)	Accessibility (1)	“Make public transport more accessible for wheelchair users, disabled, vulnerable groups”
		Capacity (4)	“Reduce occupancy of buses during peak times”, “Reduce station congestion from school drop off”
		Reliability (6)	“Improve customer confidence in public transport”
		Ticketing (2)	“Improve ticketing payment system”, “Reduce fare”
	Transport provision (18)	Match demand (5)	“Understand the ability of the system to absorb more users (long term and event based)”, “Understand demand movements”
		Alternatives to private cars (6)	“Increase public transport provision from rural areas”, “Reduce transport deserts”
		Connectivity (7)	“Improve transport connectivity”, “Social hubs interconnectivity improvements”

Figure 4. Sunburst plot of the themes (inner), codes (middle) and keywords (outer) of the outcomes generated during the workshop.

Figure 4. Sunburst plot of the themes (inner), codes (middle) and keywords (outer) of the outcomes generated during the workshop.

Theme 1 points towards evaluating mode choice and emissions from a system perspective, e.g., proportion of trips taken by mode. Theme 2 specifically relates to modal shift, seeking to evaluate the availability and attractiveness of sustainable modes to encourage their use. Understanding the desired outcomes is important, as this helps refine the Key Performance Indicators (KPIs) against which different interventions and packages of interventions should be evaluated and compared.

Figure 4 highlights that very few outcomes relate to reduced mode emissions, with attendees overwhelmingly focussed on the modal shift from private cars to sustainable modes (~95% of responses). This may reflect the bias of attendees from transport planning backgrounds. With respect to encouraging sustainable transport use, around 60% of responses relate to transport provision. Here, attendees were concerned with improving connectivity between different modes, ensuring alternatives to private cars in areas with poor mode choice and meeting current and future demand. With respect to connectivity, particularly PT provision from rural areas to urban areas to access services, covering so-called transport deserts, supporting the visitor economy and nighttime economy (evening services), and travel for medical appointments. Transport deserts are areas where a large proportion of the population rely on PT, but the available services do not adequately meet demand [29]. KPIs that could help to quantify connectivity include PT service frequency or the percentage of households within 10 min of a PT stop, differentiated by area. The remaining outcomes relate to transport user experience and comprise the different aspects highlighted as important to passengers. Almost half of these responses relate to PT reliability, indicating a particular area of concern, with a further 30% relating to capacity, e.g., bus crowding. Accessibility and ticketing were also noted.

3.3. Use Cases

Around 90 use cases were captured across Exercises 1 and 3, defined as intervention at location. Attendees were primed to consider CAP interventions reported as most appropriate for the West Midlands region. Figure 5 shows the proportion of responses within each of the five developed themes and their codes, which are categorised as physical changes to structures and changes to how structures are used. Slightly more use cases relate to changes to how structures are used, rather than physical changes.

Table 3. Thematic approach used for use generated during the workshop; brackets indicate the number of responses assigned to each theme, code and keyword.

Themes	Codes	Keywords	Example Quotations
1. Infrastructure (20)	New transport infrastructure (13)	Cycle infrastructure (2)	“Segregated infrastructure, e.g., cycle lane”
		Rail infrastructure (8)	“Local stations reopened”, “Metro across the city”
		Tram infrastructure (3)	“Tram extension to coach station”, “Tram extensions down main roads”
	Power (7)	Energy management (2)	“Energy recovery to increase efficiency of rail network”, “Load balancing”
		Electric vehicle charging (3)	“Charging of electric buses at depots/along a route”
		Rail electrification (2)	“Electrify rail lines”
2. Building and land use (18)	Land use (4)	Local services (4)	“15-min neighbourhood around work location”, “Services available more locally”
	Facilities and amenities (14)	Intermodal interfaces (6)	“Cycle parking at larger bus stops”, “Optimal location of park and ride”, “Pram, wheelchair, bike access [on trains]”
		Real-time information (3)	“[bus] stops display real-time information”, “Apps to show journey planning across modes”
		Static signage (2)	“Wayfinding—pointing to active travel options”, “Changing bus lane signage to improve understanding of road users”
		Station amenities (3)	“Stations with toilets, shops, cafes”, “Shower facilities/locker rooms available in stations”, “Family spaces”
3. Operations (23)	Transport service provision (12)	Dynamic operations (5)	“Contingency plans for strategic locations based on resilience modelling”, “Demand responsive transport”, “Disruption management”
		Route changes (2)	“Routing of buses to local centres”, “Routes—orbital routes are [too] slow”
		Additional Public Transport (PT) services (5)	“Additional buses during peak times”, “24-h bus service”
	Multimodal coordination (11)	Mobility hubs (2)	“Satellite bus centres”, “Bus hubs”
		Schedule coordination (9)	“Aligning buses and trains in rural areas”, “Optimisation of [public transport] connections”, “Integrated, multimodal public transport”
4. Traffic management (13)	Road use (11)	Bus priority (3)	“School buses to run exclusively on specific corridors”, “Priority [for buses] to pull into traffic”
		Active travel priority (3)	“Pedestrian right of way”, “School streets/walking buses”, “Pedestrian/cycle right of way in city centre”
		Restrict vehicle access (5)	“Consolidation of freight at the edge of the city centre”, “Close roads”, “City centre restrictions on personal vehicles”
	Dwell time (2)	Passenger flow (2)	“Ingress/egress on public transport”, “Buses—multiple exits”
5. Pricing (11)	Fares and ticketing (9)	Integrated ticketing (4)	“Increase smartcard area and modes”, “Integrated, multimodal ticketing”
		Free or reduced fares (5)	“Cost-ticket prices [for public transport]”, “Free public transport in low emission zone”, “Free bus travel for young people”
	Parking charges (2)	Change parking pricing (2)	“Different pricing methods for park and ride”, “Make parking more expensive for those with public transport alternatives”

illustrates how keywords were extracted from responses and then linked to codes and themes. All anonymised responses can be found in sheet 5—Use Cases (1&3)—of the Supplementary Materials Excel spreadsheet, with keywords, codes and themes assigned.

The bullet points below provide further detail on the interventions of interest and their relationship with the outcomes previously defined in Section 3.2. It is important to understand the range of interventions of interest, as this informs the data needed to create DRs.

• Infrastructure changes predominantly relate to new transport infrastructure, with responses indicating high interest in rail (62%), tram extensions (23%) and cycling (15%). This supports the outcome of transport provision. Transport power requirements are of almost equal interest for rail electrification and load balancing, and the charging of electric vehicle bus fleets, which aligns with the reduction in mode emissions.
• Building and land use includes changing land use to provide services more locally, e.g., 15 min neighbourhoods (to thereby reduce car-km) and improving transport facilities and amenities, which aligns with the outcome of improving transport user experience. Most interventions in this category (78%) relate to the latter. Around two-thirds of the responses aim to facilitate intermodal changes, predominantly between active travel and bus or rail, through the provision of cycle parking, storage, showers, etc. The remainder relate to providing better information to passengers, particularly in real time, so they have confidence to use public transport.
• Operations directly relate to transport provision and user experience. Half the responses focus on service provision, with high interest in modelling additional PT services (predominantly buses) and dynamic operations, where PT responds to changes in passenger demand and planned or unplanned disruption on the network. The other half centre multimodal coordination, predominantly by aligning schedules between trains and buses, with some mention of developing mobility (bus) hubs. There is significant weighting towards buses as an intervention, which may be due to their relative flexibility and short implementation time.
• Traffic management largely concerns road use, with 85% of responses looking to restrict vehicle access (both private cars and freight) or prioritise pedestrians and active travel specifically within the city centre. These interventions force mode shifts but also improve the safety and experience of active travel. School travel is also targeted through either school bus priority or “school streets”. Passenger flow on and off buses is also considered to improve passenger experience. A few responses highlighted the importance of motivating users to opt for active travel, e.g., by promoting health benefits, reduced cost, etc., and managing the cultural disconnect between car drivers and active travellers.
• Pricing is largely focussed on fares and ticketing (82%), with further responses related to parking charges. Free or reduced fares are of interest, specifically for young people and in low-emission zones. Integrated ticketing across modes is another intervention of interest that could improve public transport user experience to drive mode shifts. Parking charges relate to park and ride locations and large employers with private parking facilities.

3.4. Added Value of Digital Twinning

Exercise 3 collected responses to three questions on the added value of digital twinning. Attendees were asked to consider what dynamic data sources might be included and feedback mechanisms (i.e., connections to the real world) to create a digital twin for urban transport decarbonisation. Additionally, they were asked to consider how the added value could be validated. These questions encourage practical consideration about how a digital twin (defined as having automatic, bidirectional data flows between physical urban systems and their virtual representations) could be created, and critical thinking about what issues exist in traditional transport modelling and how digital twinning could overcome them. Anonymised responses can be found in sheets 6–8 of the Supplementary Materials Excel spreadsheet.

3.4.1. Dynamic or Time-Dependent Data Sources That Could Be Added to Models

Table 4. Categorised data sources for potential inclusion in an urban digital twin for transport decarbonisation.

Category	Rationale for Inclusion	Potential Sources
Active travel	To better understand walking and cycling demand, routes and experience	Smartphone/GPS, logging apps such as Strava, CCTV (e.g., at cycle parking facilities), rental bike usage data, vehicle sensing, traffic light—pedestrian interactions, accessibility data, e.g., step-free access assessment
Air quality	To understand where and when air quality is poor, and determine root causes, e.g., idling, congestion	Air quality sensors (e.g., nitrous oxides (NOx), particulate matter (PM), sulphur dioxide (SO2), carbon oxides (COx))
Parking	To understand car park utilisation and how long people are using services	Car park ticket barrier data, car park occupancy sensors, car park ticketing, park and ride ticket data, CCTV
Power and energy	To understand power and energy demand from different substations at different times from public and private transport	Train traction charging, train metering data, electric bus charging demand from depots, electric vehicle charging data
Public transport occupancy	To understand differences in demand across routes and at different times (peak/off-peak, through the week, seasonal variation, for specific events). To improve the accuracy of metrics such as emissions per passenger	Smartphone/GPS, CCTV (e.g., station platforms, concourses, bus stops), occupancy data from operators, ticketing data
Road traffic	To understand the difference in demand across routes at different times	Smartphone/GPS, traffic sensors, road conditions data, low emission zone data including usage by compliant and non-compliant vehicles, car count data, traffic management protocols, routing protocols, e.g., from Google Maps, road traffic control centre data, CCTV, automatic number plate recognition
Social media	To understand real-time issues with services and passenger sentiments towards PT	Different social media platforms, e.g., X, Facebook, Instagram
Transport-related social exclusion	To identify people experiencing exclusion to ensure interventions alleviate their issues. Contributors to exclusion include poor provision of local transport, unsuitable conditions to facilitate travel, high level of car dependency. It is most likely to affect people with disabilities, health conditions, caring responsibilities.	Census data, demographic information, accessibility information, information on street conditions, local transport schedules, requests for passenger assistance
Journey demand	To understand demand and opportunities for modal shift, including for specific events	Smartphone/GPS, travel survey data from employers, boarding/ticketing data from operators, smartcard data, data from ticket barriers, ticket data for special events, e.g., concerts, sports matches
Weather and environment	To understand the impact of these conditions on transport, e.g., mode choice, potential disruption	Weather sensors, Met Office data, climatological data

captures the different categories of data that attendees identified, alongside a rationale for their inclusion and a list of potential sources—noting that the same source could be used to derive multiple insights, e.g., smartphone/GPS for travel demand and travel on individual modes. It is assumed that these sources are not typically included in existing models and therefore represent added value. However, even where reliable, real-time sources exist, it may be difficult to access the data.

3.4.2. Feedback Mechanisms Between the Real World and DR

Only 18 responses were received with respect to feedback mechanisms, indicating that these may be limited or that the understanding of what is possible is limited. They broadly relate to customer support with journey planning, providing real-time updates during a journey and operational control. They also predominantly rely on human interpretation of real-time conditions to make informed decisions. Such connections would facilitate interaction and control to improve day-to-day urban transport operations but are less relevant for longer-term scenario modelling. However, control over a longer period becomes more possible if the real-world transport system replicates the digital world rather than the other way around, e.g., through automatic train control, proliferation of autonomous vehicles, etc.

• Journey planning
  • Assumed to be through online planners or mobile apps;
  • Providing information to passengers on alternatives to car travel, including pricing;
  • Sharing the availability of parking spaces, e.g., disabled spaces, electric vehicle charging spaces, parent and child spaces; status of electric vehicle chargers.
• Real-time updates
  • Customer information systems in railway stations, on trains and via apps, providing useful data to travellers such as the location of the quiet carriage, empty luggage racks, seat availability, for their journey;
  • Customer information systems at bus stops, on buses and via apps, providing updates on service occupancy, location, expected arrival, etc.;
  • Update to traffic conditions via routing apps.
• Operational control
  • Smart motorway gantry control;
  • Sensor-based traffic light control;
  • Information provided to traffic controllers, road and rail (e.g., signallers, platform dispatchers) and transport operators.

3.4.3. Validating the Added Value of Digital Twinning Methods

The intention of this exercise was to determine specific interventions where traditional and digital twinning methods could be compared to validate the added value. Although some specific projects were mentioned, the majority of the 16 responses identified current limitations more generally, indicating that digital twins can add value where traditional methods fall short. A digital twinning approach for evaluating transport decarbonisation interventions should aim to overcome limitations by:

• Identifying the system effects of changes, rather than just primary relationships;
• Using data feeds to give more specific (temporal) parameter values, rather than averages;
• Incorporating variance and elasticities, including human variance;
• Validating interventions found using human factors techniques within the models;
• Evidencing the correctness of models against the real world;
• Allowing users to visualise the impact of changes;
• Enhancing usability to encourage the use of models and provide greater access.

3.5. Framework of Five Critical Aspects to Inform Digital Twinning Use Cases

Figure 6 summarises the impacts, interventions, location types, data sources and feedback mechanisms of interest captured during the workshop in relation to the generalised digital representation presented in Figure 2.

• Services: the outputs captured in the workshop support the specification of the transport interventions to test, the types of locations to which they should be applied, and relevant decarbonisation metrics to support local transport planning and project evaluation.
• Data sources: by incorporating the data sources captured during the workshop, modellers can add value, creating a more realistic representation of the real-world transport system (including travellers) and the wider system within which it sits. Baseline data is also required to be able to provide the service functionality desired by users. As previously stated, this is true for models, shadows and twins.
• Feedback mechanisms: the mechanisms captured relate to more automated day-to-day operational control and information provision to drive behavioural change. For longer-term scenario modelling, the insights require human interpretation and will inform decision making.

The different aspects of the framework are described in

Table 5. Description of what each aspect adds to the framework for digital twinning use case definition.

Aspect	Description
Impacts	Captures the impacts desired from urban transport decarbonisation interventions. From this, appropriate KPIs can be defined to compare different interventions or sets of interventions.
Location type	Specifies the areas of interest to be modelled, helping to define the system boundary and scale of the DRs required.
Interventions	Sets out the transport decarbonisation interventions of interest. This helps to specify the level of detail required for the DRs and provides a starting point for baseline development.
Data sources	Prompts developers to consider additional dynamic data sources of value to be incorporated into the DRs to create a more realistic representation of the real world.
Feedback mechanisms	Prompts developers to consider how the impact evaluation and insights derived from the DRs can be fed back to influence the real-world state of the transport system. Direct control is achieved through a DT.

. Application of the framework encourages DR creators to explicitly consider the services to be provided to users and how the derived insights influence the real world and the data connections between the physical and digital, noting these are often overlooked in reported research.

3.6. Illustrative Example of Framework Application

We applied the developed framework to define five digital twinning use cases for a prototype multimodal passenger transport DR of the West Midlands, with one use case selected for trialling based on stakeholder appraisal of various selection criteria. To illustrate this without detailing all the use cases and processes in full, we set out an example relating to managing travel to and from Edgbaston Stadium to reduce the accumulated carbon emissions of event days. Spectator travel contributes as much as 80% of the total match day emissions [30].

Edgbaston Stadium has a capacity of 25,000, with cricket match-day attendees travelling in locally from Birmingham, more widely from locations around the UK, and internationally. It is located south-west of Birmingham city centre on Edgbaston Road, close to a main arterial road (the A38), and is approximately 2 miles away from three railway stations on the cross-city commuter line. When Moseley station reopens in 2026, this will be 1 mile away. The stadium is served by local buses, with paid shuttle buses running from the city centre on match days. There is limited parking available at the stadium that must be prebooked. Road closures are put in place for busier match days, affecting local traffic.

Table 6. Illustrative example of framework application.

Aspect	Description for Edgbaston Stadium Use Case
Impacts	This use case considers how to manage travel to and from Edgbaston Stadium for cricket matches through improved transport provision and transport user experience and to understand the impact on local travel. The desired outcome is a modal shift from private cars to public and active transport modes to reduce carbon emissions from spectator travel. Example KPIs: mode share, CO2 per spectator-trip, PT capacity, PT occupancy, park and ride usage, road traffic congestion (e.g., average delay or travel time index).
Location type	Transport desert—although there are frequent bus services to and from the city centre to the stadium, the stadium is not well connected to locations within Edgbaston and other areas of Birmingham. Railway stations (on commuter lines) are at least a half-hour walk away. It is next to an arterial road, which may be affected by local road closures on match days.
Interventions	Interventions predominantly relate to transport service provision. As a special case of dynamic operations, contingency plans for match day travel could be developed. These may include additional PT services, route changes or park and ride from alternative stations to Birmingham New Street in the city centre. Given that road use changes (i.e., access restrictions) are already in place, revisions to these could be evaluated.
Data sources	Understanding journey demand is crucial, including the modes (e.g., active travel, PT) by which visitors travel to the stadium. To understand the impacts this has on congestion and residents, data is needed to compare average congestion vs. match day congestion with regard to parking, PT occupancy and road traffic.
Feedback mechanisms	Feedback is likely to concern journey planning or real-time information for match day visitors or residents affected by road closures and traffic.

details the different aspects to be considered for the Edgbaston Stadium use case. It should be noted that the framework is intended as a first step in DR design and development, and further work is necessary to fully conceptualise the representation(s) to be developed from a technical perspective.

Figure 7 highlights relationships between aspects for the Edgbaston Stadium use case. The proposed DR would deliver the service “scenario selection”, i.e., scenario modelling capability to transport planning users. Real-world data sources, from a range of relevant locations, feed a baseline model and inform the simulation of interventions. The impact of interventions is to be assessed using a range of KPIs (against the baseline), to support decision making regarding which interventions to implement. Behavioural changes could be driven through feedback mechanisms, such as journey planning and real-time information provided to spectators. These changes would be captured through ongoing, dynamic data collection. Note that in a digital twinning ecosystem, this DR would be complemented by others that provide different services relevant to the use case, developed from overlapping data sources, for example, a digital shadow to support operators in monitoring congestion on match days, or progressing from this, a DT to directly control road traffic (e.g., through traffic signals, road signage).

4. Discussion

4.1. Critical Evaluation of Contributions

Whilst much of what is presented here with respect to the impacts, interventions and location types (and to some extent, the data sources and feedback mechanisms) of importance for urban transport decarbonisation is known, few, if any, studies directly capture and present practitioner perspectives on these aspects. It is possible that presenting the discussion in relation to the CAP may have limited the interventions generated to current solutions, rather than challenging attendees to consider more radical interventions. However, this was done deliberately as most attendees would have working knowledge of the CAP, providing a solid foundation from which to begin the discussion.

The five aspects of the framework proposed in this study were selected to enable practitioners involved in the development of urban digital twinning solutions to maximise the value generated during limited discussion time with busy stakeholders in the key criteria of the urban planning process, using requirements they can understand, namely, desired outcomes (impacts, locations), solutions that could be employed (interventions), and the opportunity for measurement, refinement, and evaluation (data sources, feedback mechanisms). The alignment of the framework aspects with the full set of artefacts described by Nag et al. [14] in their definition of a “digital twin” ensures that the use cases being captured meet the criteria for, and have the potential to be practically implemented as, full digital twinning solutions. In the wider industrial landscape, standard frameworks for the establishment of DTs exist and serve as a basis for many of the national guidelines (e.g., the outputs for the UK NDTP). Commonly used examples include the ISO 23247 series (representing a full workflow in a manufacturing environment, including general principles, reference architecture, data representation, and information exchanges), the ISO 19650 series (focussed on Building Information Modelling (BIM)), and ISO/IEC 30172 (focussed on use cases). Many of these activities are focussed on the technical aspects of delivering the DT, typically in a physically constrained context, such as the confines of a manufacturing plant; here, the model leads the use case, with the entire asset base within the physical system boundary being captured. The geographically dispersed nature of the urban environment does not lend itself to the same level of in-depth modelling, and so here, the proposed framework complements technical modelling frameworks by providing a focus on the choice of modelled assets, first by location, and later by function. In the case of standards such as ISO 30172, it does focus on the capture of use cases, but in the context of the Internet of Things (IoT), the complementarity of the proposed framework comes in the structuring of the case study around specific interventions being introduced into the system; here, the feedback loop is long, and a “right time” response must be delivered in contrast to the “real time” response typical of the IoT.

We recognise that not all the impacts pertinent to urban planning are captured, particularly those related to an individual’s experience of transport and the potential wider social benefits of transport interventions, such as increased productivity, job creation, and health benefits. There are likely additional aspects related to transport user experience that were not explicitly captured, although our attendees brought knowledge gained from engaging directly with user groups. We also recognise that it is important that the impacts of interventions are assessed in terms of cost and benefits; one response captured this as “understanding of capital versus revenue expenditure benefit”, but generally, the costs of interventions (to transport authorities, operators, maintainers, etc.) were not considered. There is further work and engagement to be done to comprehensively catalogue the potential impacts of transport interventions and define KPIs that enable users to evaluate different solutions.

4.2. Critical Evaluation of Approach

The workshop attendees were all mid/senior-level professionals in academia and industry, which, as discussed above, may have limited the responses captured. To better understand the transport user experience of individuals in future, we should engage with communities, including representatives from vulnerable populations. This should include both those who are least elastic in response due to their needs and those most impacted by the proposals.

The workshop was designed so that attendees generated data directly by writing responses to questions. However, attendees discussed these questions in groups, raising various issues that may not have been fully captured. To generate deeper insights, it may have been beneficial to record the discussions of each group during exercises 1–3 or allocate a scribe to each table to capture salient points. This would have provided an additional source of data for analysis. This approach will be considered for future workshops.

We decided to employ reflexive thematic analysis, targeting a nuanced interpretation of the data gathered and flexibility in the coding process. Rigour was ensured by following Naeem et al.’s systematic approach [26] with transparent reporting. The developed codes could have supported the development of a coding scheme to be used by two or more independent coders, enabling inter-coding reliability methods to be applied [31]. However, project resource and time constraints meant that this was not feasible. By including multiple researchers in the affinity diagramming (including non-workshop attendees) and by checking the results, we sought to reduce potential bias.

4.3. Digital Twins vs. Digital Twinning

One question raised by the current work is whether a DT is necessary to evaluate strategic transport decarbonisation solutions. Taking inputs from the physical world to create a DR is critical regardless of whether a model, digital shadow or DT is being built, and the identified data sources could arguably add value to any level of representation. Automated feedback mechanisms (that enable the representation to directly influence the real world) are the differentiator between a DT and shadows or models. However, the limited responses provided by practitioners in reference to feedback mechanisms suggest that (under current circumstances) the influence of interaction and control is predominantly on day-to-day operations. These temporary measures will undoubtedly have a much smaller impact on the net-zero transition than longer-term strategic decisions, such as building a new railway or mandated adoption of electric vehicles. Yet implementing these longer-term decisions requires human intervention, so is the tool used for evaluation no longer a DT? Or is data flow via humans a valid feedback mechanism to influence the real world within a DT? Though a question of semantics, this is why we introduced the concept of digital twinning.

TransiT is a hub for digital twinning focussed on federating DRs across transport modes, which should be at the appropriate level of fidelity to support the evaluation of decarbonisation solutions. The ecosystem is expected to include models, shadows and DTs as appropriate for different functions. The digital twinning components will be designed to meet articulated challenges, with DTs utilised where appropriate for operational decision support and models of future infrastructure and demand for longer-term planning. We will continue to develop our DR(s) of multimodal passenger transport in the West Midlands UK, to explore our chosen use case(s). To progress from the framework, further work is needed to design and develop the required DR(s), first as a blueprint and then technically. Initial work includes gathering and evaluating data sources, generating evidence to help specify the interventions in greater detail, identifying tool users and stakeholders, and working with them to define specific metrics that would support them in evaluating the impacts of interventions. As development progresses, we will further work to embed co-creation in the project to ensure utility for end users, drawing on the expertise of stakeholders within the TransiT research hub. Co-creation extends beyond engagement, requiring traditional phased approaches to development projects to incorporate new methods, tools, competencies and management rules that facilitate user integration [32]. When planning the development lifecycle, thought should be given to the tools and methods used to engage different users in co-creation subphases, including design, testing, launch and production. Our aim is to work towards delivering the added value functionality identified by transport planning practitioners, namely the ability to conduct multimodal what-if analysis, generation of real-time insights, inclusion of realistic passenger behaviours, and improved tool quality and robust data.

5. Conclusions

Through this research, we sought to capture practitioner perspectives on the specific functional uses for an urban-scale DT for transport decarbonisation. We have answered the research questions: which impacts, interventions and location types are seen as priority concerns for transport decarbonisation, and which data sources and feedback mechanisms are viewed as value-adding contributions (as stated in Section 1.4)? Practitioners are overwhelmingly interested in evaluating mode choice, with 95% of outcomes related to increased sustainable transport use occurring through better transport provision and user experience. Responses indicate almost equal interest in assessing physical changes to transport structures and changes to how they are used. Priorities are new infrastructure, improved facilities, additional PT services, multimodal coordination and road use changes, particularly to connect transport deserts. A wide range of data sources could be incorporated to improve digital representations, particularly smartphone/GPS data. However, the feedback mechanisms necessary to create a DT are limited to mobile applications, customer information systems and traffic control. With reference to the third question, we have used this information to develop a framework with which practical digital twinning use cases can be defined.

By documenting expectations from transport system digital twinning captured directly from practitioners, this paper addresses a gap in the literature (as identified by Nag et al. [14]). We hope the methods, data analysis and framework outlined in this paper will be of use to other researchers in developing digital twinning applications for urban transport decarbonisation and in other sectors, with due consideration of user needs. The framework helps to centre development around the aspects identified as critical to users—namely the services provided by the tool and the digital twinning data sources and feedback mechanisms that add value over existing approaches. It provides the first step in the development of a digital twinning ecosystem for transport decarbonisation. Developing multiple use cases is necessary to understand and capture the range of functionality desired from a single platform.