A user-driven design approach towards developing a citizen-centric digital twin for maintaining urban infrastructure Open Access

Digital platforms allow citizens to get involved in rectifying social issues and city-related concerns while improving their societal needs (Testi et al., 2025). A citizen-centric digital twin (CCDT), as suggested by Abdeen et al. (2023), can obtain citizen input and process it in real-time to identify issues and problems relevant to infrastructure and other societal needs. This data-driven approach ensures that governments’ investments are more accurately aligned with infrastructure and societal needs and can lead to a more efficient and effective allocation of resources (Townley and Koop, 2024). Furthermore, leveraging technology and citizen participation plays a major role in planning and managing urban environments and the governing system of smart and sustainable cities (Alamoudi et al. (2024). Similarly, Alkhalifa (2024) identified that digitalization, the level of citizen engagement and the efficiency of city infrastructure are considered key factors for smart sustainable cities. The study attempts to address this integration using CCDT by taking a more citizen-centred approach to smart sustainable cities.

As one of the use cases, CCDT can be used to solve critical issues related to transport infrastructure maintenance. According to Gutteridge Haskins and Davey (2015), the lack of data is one of the major concerns in achieving continuous transport monitoring and assessment of maintenance needs. The report further states that maintenance backlogs have resulted in an estimated cost for potential road accidents of $AU27 billion per annum in Australia. Do et al. (2024) stated that the current approach for infrastructure condition monitoring is centralised within the public authorities and solely depends on the technical expertise and assessment of the relevant officers. These systems lack public engagement within the localised infrastructure system, which could lead to inaccurate judgements of the infrastructure conditions (Do et al., 2024). CCDT offers a solution to overcome this gap while providing access to more in-depth and contextual data collected using citizen inputs to provide a data-driven approach to transport infrastructure maintenance. As the International Transport Forum (2021) states, a more data-driven approach to transport infrastructure maintenance improves the quality of assets, enhances their life cycles and saves costs incurred during maintenance. Despite the offered benefits, designing and utilising an engagement platform for public participation has its challenges (Karkin and Cezar, 2024).

Between 60% and 80% of software project failures can be directly attributed to poor requirements gathering, analysis and management (Bellisoy, 2019). Chan et al. (2025) state that citizens are rarely involved in designing civic participation, and their expectations are not clearly understood, even though they are the main users. According to Maas et al. (2021), failure to use most of the civic participation tools lacks the analysis of the customer requirements (CRs) and motivation to engage. Hence, CRs need to be carefully examined and translated into design elements to develop a customer-centric product (Fabio et al., 2025). Quality function deployment (QFD) provides a systematic approach to transferring CRs into engineering characteristics (ECs) to increase customer satisfaction (Fabio et al., 2025). As stated by Park et al. (2025), QFD has been widely applied in technological research, as it supports the development of user-centric systems. The QFD method deploys the voice of the customer in searching for the best solutions for the design and development (Singhaputtangkul, 2017) of the CCDT platform. This study, in particular, seeks to identify the ECs for a set of CRs identified for CCDT, which is one of the key stages for QFD.

Recent research in integrating QFD into digital twin technologies to understand ECs is mainly focused on the built environment, facility or organisation levels. Some examples include the work of Liu et al. (2023), where the study focused on introducing a methodology for selecting digital twins for organisations based on the quality target and resources. The study of Himmelstob et al. (2023) used QFD to design a digital twin for an asset administration shell for a component manufacturer. Using QFD, Newrzella et al. (2022) designed a methodology to determine the most impactful digital twin use cases at an organisation requiring low effort and high scalability. The focus of Himmelstob et al. (2023), Liu et al. (2023) and Newrzella et al. (2022) has been more on a controlled environment with specific activities and involves qualified personnel with a technical understanding of digital twin operations. These digital twins are designed to serve organisational requirements and are not open to the public, and public engagement has not been the focus. Therefore, it is evident that there is a lack of empirical studies exploring the ECs for a CCDT. This hinders the development of a CCDT due to a lack of understanding of the design features, such as ECs that meet the user’s needs. Ryseff et al. (2024) on the root cause of failing AI projects state that, by estimates, 80% of AI projects fail due to a lack of robust engineering and technical requirements, with a limited understanding of how to transform desires into action. A similar impact on CCDT, which incorporates AI for critical elements such as enhanced simulation, prediction and autonomous operations, could be witnessed if critical ECs are not investigated. Therefore, this study seeks to develop a framework that assists in determining the ECs based on the user needs of a CCDT. The study addresses the research question, “How can the ECs of a citizen-centric digital twin be framed during the design of a CCDT?”. A framework is developed by considering various ECs required for a CCDT and is categorised and organised into hierarchical levels.

This study contributes to the body of knowledge in several ways. Firstly, the study introduces a novel framework for determining the ECs of a CCDT by considering user requirements. The past studies derived ECs for digital twins at organisation levels, for specific activities and involved a controlled environment with digital twins accessed by technical people. Examples of a few such studies include Himmelstob et al. (2023), Liu et al. (2023), Newrzella et al. (2022) and Tang et al. (2024). Furthermore, Liu et al. (2016) pointed out that the design requirements of digital twins have extensively focused on the digital twining characteristics such as a physical object, a digital object and the connection. However, CCDT allows public involvement and functions in an open environment and hence considering other user needs to facilitate engagement is vital. As detailed in Section 4, the framework proposed by the study incorporates distinctive characteristics such as ECs to govern and manage citizens' inputs, such as comments and flagged issues. Therefore, the framework proposed by the current study, while considering digital twin-specific features, also incorporates critical ECs required for the digital twin to operate in an open atmosphere with citizen involvement. Secondly, the study added theoretical significance by improving the QFD method to suit the user-centred design of CCDT. To do so, the study utilised principles for spatially enabled digital twins of the built and natural environment to frame the ECs and use the spectrum of public participation in devising CRs.

The use of digital platforms provides the opportunity to gather collective intelligence and for citizens to be involved in rectifying city concerns while improving the efficiency of city governance (Arana Catania et al., 2021). CCDT includes infrastructure, human dynamics, spatial and temporal information flow and physical and virtual connectivity capability (Francisco et al., 2020), which is open for the public and citizens to collaborate. Citizens can update this platform on a timely basis, envisioned to improve monitoring, control and decision-making through enhanced visualisation and interaction. Integrating features and functions for citizen engagement, the digital twin is developed to be citizen-centric for more informed decision-making and with the potential to lead towards open transport infrastructure maintenance systems (Abdeen et al., 2023). Past studies on the civic digital twin by Luca et al. (2024) incorporate a citizen-centric approach to urban planning and governance. Authors mainly define civic digital twin and introduce the representation model and the research challenges. The work, however, lacks the detailed design requirements of the digital twin model and how user needs are integrated. White et al. (2021) developed an interface for reporting issues in the city in a digital twin environment. These reports are then sent to city councils to address the issues. However, the study lacks the investigation of user needs, thereby questioning the appropriateness of the considered design features for a CCDT. CCDT is an emerging concept, yet the design features or ECs that ensure the development of a CCDT by considering both digital twinning features and features to enable citizen engagement are lacking.

To understand the design features of CCDT, QFD offers a systematic approach to converting CRs and expectations into ECs to enhance customer satisfaction (Akbas and Bilgen, 2017; Wu et al., 2021). QFD originated in the manufacturing industry to integrate CRs in the product development stage, resulting in saving large amounts of costs (Han et al., 2019). Akbas and Bilgen (2017) found that the literature on QFD has seen a gradual rise since the breakthrough in Japan between the 1960 and 1970s and has since been used in various disciplines ranging from manufacturing to service and education. The theory has contributed to the computer science field with a focus on developing mobile applications (Lim and Chin, 2023) and software development (Gu et al., 2019). In the case of ICT (Sopegno et al., 2016) and smart technologies (Kamvysi et al., 2024), QFD is recognised as a tool in creating new systems. Therefore, the current study is based on the QFD method to identify the best solutions for the design and development of a CCDT platform using CRs.

In classical QFD, the House of Quality (HOQ) is a principal and strategic tool to translate CRs into related ECs (Lim and Chin, 2023). HOQ is a framework that includes all feature metrics required for a QFD structure. The developed HOQ for a product includes information enabling detailed analysis of the product development. Generally, an HOQ is constructed with seven elements: CRs, ECs, the importance of CRs, the interrelations between CRs and ECs, the competitive assessment matrix, the inner dependence among ECs and the overall priorities of ECs (Wu et al., 2021).

Several studies utilise QFD to develop solutions in the digital space or digital platforms. For example, Kirbac and Ergenc (2021) integrated blockchain for effective digital supply chain management. The authors determined the CRs of service companies and food retailers along the supply chain and the technical specification for blockchain technology under the QFD method. Similarly, Li and Kim (2023) used QFD to translate user experience into the design requirements of cultural heritage apps. Based on the design requirements, the authors developed a user experience design framework. Fang et al. (2023) used QFD to optimise the design method for an elderly medication reminder mobile application. Based on prioritised designs, four prototypes have been proposed for the app. Bahauddin et al. (2024) used QFD to increase the quality of mobile banking services by transferring CRs to technical responses. Finally, the authors suggested strategies to enhance the user experience. Kamvysi et al. (2024) used QFD to identify smart technologies when designing and implementing smart city strategies. Hence, it is evident that QFD is widely applied within the technological sphere to develop customer-centred products.

QFD in digital twin applications is noticeable in recent efforts where the applications are dominant in the built environment, facility or organisation levels. Examples include the work of Liu et al. (2023), where the authors introduced a methodology for choosing digital twins for facilities based on quality targets and resources. Similarly, Newrzella et al. (2022) utilised QFD and developed a method to understand different digital twin use cases with low effort and high scalability for an organisation. Furthermore, QFD has been used to design a digital twin for an asset administration shell for a component manufacturer by Himmelstob et al. (2023). Tang et al. (2024) used QFD to digital twin-based Prognostics and Health Management systems to convert technical requirements to design requirements. While the studies of Himmelstob et al. (2023), Liu et al. (2023), Newrzella et al. (2022) and Tang et al. (2024) focus on digital twins, these developments are focused at the organisation level, and specific activities involve a controlled environment with digital twins accessed by technical people. Furthermore, the studies have not identified or prioritised CRs and ECs focusing on citizen engagement. Moreover, Liu et al. (2016) pointed out that the design requirements of digital twins reflect digital twining characteristics, such as a physical object, a digital object and the connection. Despite numerous studies utilising QFD, few studies thoroughly investigate the potential of digital twining to foster citizen engagement, taking into account the CRs and the ECS that facilitate user needs. The current study takes a novel approach to overcome the above-discussed gap in the literature and extend the QFD method for digital twins and citizen engagement contexts. This study integrates the spectrum of public participation and principles for spatially enabled digital twins to derive CRs and ECs, respectively.

The first step of the HOQ, which is the principal and strategic tool of QFD, is to derive the CRs of the platform. However, contextualising the platform goals into the QFD method is paramount to derive CRs that create opportunities for citizen engagement, leading to the development of a digital twin with engagement capabilities. Therefore, the CRs derived from the study should be relevant in meeting engagement requirements. To achieve this, the Spectrum of Public Participation (SPP) developed by the International Association for Public Participation was used (Doğu et al., 2024). The CRs were identified in alignment with SPP categories to ensure the CRs reflect the engagement activities expected from the CCDT.

The SPP comprises five categories: inform, consult, involve, collaborate [with] and empower the public (Evans et al., 2025). Inform refers to the role of the participation tool to provide balanced information to the public. Consult refers to the role played by participation in obtaining public feedback. Involve relates to the ability to work directly with the public to understand their concerns. Collaboration provides opportunities to partner with the public in decision-making, whereas empowerment allows the public to take a place in final decision-making (Evans et al., 2025). According to Doğu et al. (2024), SPP assists in reviewing the public’s role through the participation process using these five stages. Additionally, SPP provides a realistic depiction of participation and engagement activities and serves as a benchmark and a reference for classifying engagement activities and evaluating public involvement and engagement initiatives. Hence, SPP suits the study’s requirements in understanding the CR of CCDT. Table 1.

The next stage of the HOQ is to identify the ECs. Ensuring that the identified ECs reflect the digital characteristics is vital to developing the CCDT. As HOQ only provides an overall framework, customising it to ensure the specific context of the platform is necessary to obtain valid results.

Therefore, the principles for spatially enabled digital twins of the built and natural environment (The Australia and New Zealand Land Information Council (Anzlic), 2019) were considered in this study in deriving the ECs through the literature review. These principles describe the overall intent of the spatially enabled digital twins, the purpose and key elements under each principle. The nine key principles are public good, standards, quality, openness, security and privacy, curation, adaptation, value and federated model. Identifying the ECs along with these principles ensures that the developed CCDT reflect spatial digital twin characteristics. A spatial digital twin is capable of capturing 3D and 4D spatial data (referencing a specific geographic location) to model the urban environment (Commonwealth Scientific and Industrial Research Organisation, 2023). This capability to capture spatial data on top of nonspatial data is critical for developing digital twins at the city scale in the context of citizen engagement to provide accurate information for citizens. Hence, the study considered principles for spatially enabled digital twins of the built and natural environment in identifying the ECs (Table 2). The principles draw on the UK’s Centre for Digital Built Britain’s Gemini principles, a framework for developing and implementing digital twins. This will ensure that the CCDT designed by the current study meets the key principles of spatial digital twins and aligns with the digital twin frameworks.

Figure 1 depicts the research method flowchart of the study deployed to collect and analyse data under qualitative research methods. The content of Figure 1 is explained in the following sections.

The qualitative interviews provide an in-depth understanding of the ECs related to CR while obtaining sufficient clarification and justification. Hence, the study was based on qualitative interviews. The study sample was selected based on purposive non-probability sampling. Unlike probability sampling, which focuses on generalisation across a broad population, this study focused on obtaining an in-depth understanding of how ECs should be framed during the design of a CCDT. Therefore, using purposive sampling, the study participants were intentionally chosen based on their ability to provide rich and meaningful insights on framing ECs for CCDT. This also assisted in overcoming limitations due to the lack of experts on digital twins and allowed the researchers to identify a valid sample with digital twin expertise and improve the trustworthiness of the data. The study considered those who meet the following criteria: (1) Have been in practice for a minimum of three years in the areas of geographic information systems or spatial digital twin or city information modelling; (2) Have been involved in projects on a city scale; and (3) Knowledgeable of the system architecture of digital twins. Participants were approached by circulating the recruitment flyer through the details available on the websites and were requested to contact the researcher if they met the recruitment criteria. Informed consent was obtained before the interview while maintaining participant confidentiality. In qualitative interviews, the number of participants or sample size could vary based on several factors, such as time, resources and the objective of the study (Agarwal et al., 2020). This study involved 20 semi-structured interviews with a sample size determined based on the principle of data saturation. In this study, data saturation was observed after the 18th interview, with additional interviews carried out no longer yielding new themes or generating new insights. Prior research suggests a sample between 16 and 24 is sufficient to reach data saturation (Hennink et al., 2017). Due to the homogenous nature of the participant group and in-depth interviews conducted, 20 participants provided comprehensive insights required to answer the research questions of this study.

The semi-structured interview guideline was developed using the CRs and ECs identified in Tables 1 and 2, respectively. The respondents were required to identify whether existing ECs in the interview guide are relevant to meet CRs, and if not, propose additional ECs to meet the CRs. During the interview, respondents were asked their opinion on considering ECs identified from the literature and the reasons for the proposed ECS. The semi-structured interview guide was designed for interactive and engaging sessions using Mural, an open and freely available platform. The Murals were shared among the participants before the interview to help them familiarise themselves with the process. The Mural provided supportive materials, including identified ECs from the literature. Participants were guided to use supportive material or propose additional ECs based on suitability. The use of Mural allowed participants to understand the structure of the interview, and it also presented the connection between CRs and ECs. This ensured the overall understandability of the interview process. Also, participants were able to interact with Mural and write their thoughts on it to ensure they stayed on track and focused. This allowed the researchers to seek justification, clarification and cross-question the experts. In some cases, Mural could offer drawbacks due to lagging, poor Internet or overcrowding. However, during the interview, no such issues were experienced.

In this study, the content validity of the data collection instruments is ensured through validation interviews. Yusoff (2019), by considering several other studies, proposes that the recommended minimum number of experts for content validation is six. Based on the findings of Yusoff (2019), six experts were considered sufficient, as the generated feedback from the experts was sufficient to identify issues related to excessive or redundant questions. Participants representing the disciplines of Geographic Information System (GIS), digital twin and city administration or transport infrastructure management were considered for validation interviews. These participants were required to have knowledge and experience of user-centric design to ensure the interview structure was understandable. Participants who were willing to provide constructive feedback were considered. To ensure the reliability and validity of the findings, strategies proposed by Bashir et al. (2008) were used such as (1) participants' language and verbatim were considered, literal statements from the participants were considered in the analysis process, (2) Negative or discrepant data/statements were considered and (3) the interviews were recorded.

As depicted in Figure 1, data analysis was performed using inductive and deductive coding techniques. During the analysis process, three methods were used to ensure the validity: (1) generating codes manually and using NVivo, a qualitative data analysis software, (2) using deductive and inductive codes and (3) utilising comparative analysis to identify codes/themes.

Initially, deductive coding using content analysis was carried out to understand the applicability of ECs from current literature (Table 2) for CCDT development. ECs identified by participants were categorised based on the engagement levels of inform, consult, involve, collaborate [with] and empower. The identified ECs were mapped against the ECs identified from the literature review. Similar to the studies of Wang and Chen (2020), the average percentage of agreement of each of these ECs was calculated to identify ECs with high levels of agreement. This allowed the researchers to identify critical ECs, derive patterns and identify the themes. ECs with an average percentage of agreement above 15% were considered, as most of the ECs identified fell within the agreement rates of 15%. ECs not meeting this threshold were considered to be of low average percentage agreement. This ensures that the ECS identified remains relevant and prominent in a digital twin setting to promote engagement. Microsoft Excel was mainly used when calculating and synthesising data during the inductive coding.

Next, the inductive coding using thematic analysis was carried out to identify additional ECs that are relevant to CRs (Section 4.2). The additional ECs are the ECs that emerged during the interview, in addition to the ones identified from the literature. During inductive coding, opening codes were assigned to capture initial ideas. These codes were refined and grouped to better represent the idea. Final themes were then generated and reviewed. In the final stage of the analysis, comparative analysis (Section 4.2) was carried out. During the comparative analysis, additional ECs identified by single participants were manually mapped against the transcripts of all other participants. Implicit ideas, comments, examples and justification were considered, and the results were mapped. Adopting multiple analysis frameworks enabled the researcher to derive valid and robust findings. NVivo was used during the inductive coding and comparative analysis process. Finally, the themes derived using the deductive and inductive coding were integrated to propose a framework for deriving ECs of a CCDT.

The section presents the results of content and thematic analysis carried out using a deductive coding process. To perform the content analysis, the average percentage of agreement of ECs within the engagement levels of inform, consult, involve, collaborate and empower was calculated. Table 3 represents the ECs classified based on the average agreement rates by participants (AGRP) above 15%. The ECs retrieved from the literature and highly referred to by participants were explored to identify possible themes. This process enabled the identification of initial themes of ECs that are prominent in meeting the CR of a CCDT. The findings of this section are integrated with Section 4.2 to address the overall purpose of this study.

The analysis in Table 3 shows that EC4 and EC16, which are user-friendly and easy-to-navigate graphical user interfaces and workflow automation, are two commonly reported ECs to meet CR across all five categories. Fifteen among the 22 identified ECs from the literature highly support CRs focused on informing the citizens, consulting the citizens, involving them and empowering them, with an average percentage of agreement between 15% and 53%. Hence, the literature is rich with ECs that support citizens' involvement and consultation. However, the analysis in Table 3 revealed that only EC4 and EC16 within the category for collaboration received an average percentage agreement rate above 15%. Hence, there is a noticeable gap in the literature on ECs that allows citizens to collaborate within the platform. Based on the identified ECs from the analysis, key categories were derived and explained below.

Among the ECs identified from the analysis in Table 3, EC4, which is the user-friendly and easy-to-navigate graphical user interface, received the highest agreement rates of 0.33, 0.38, 0.29, 0.53 and 0.26 to enable features to empower, collaborate, involve, consult and inform, respectively. This emphasises the need for EC4 across all five categories of public participation within the digital twin. EC4 was considered to highly support the consultation process of the citizens, with more than 50% average percentage of agreement. According to P14, EC4 benefits Novis users of the digital twins as

Preserving the security of the citizens is a key theme that has emerged and facilitates the reliability and sustainability of the CCDT. EC11, EC14, EC13, EC6 and EC2 were identified as predominant ECS within the category. The average rate of agreement of these ECS among the participants ranged from 15 to 22%. Participant P2 stated

As the analysis revealed, all 5 ECs identified within the theme were required to enable features to empower and involve the citizens. Therefore, the ECs within the theme were mostly related to empowering and involving the citizens within CCDT.

Governance and data quality are the third themes generated during the analysis. This theme mainly focuses on having clear ownership to support governance and ongoing developments and meet minimum data quality standards with an agreement rate of 20% and 17%, respectively. The ECs were mainly enabling CRs that empower and consult citizens. Several key areas to be addressed to ensure governance and data quality were discussed by participants, as the CRs focus on gaining citizen inputs for informed decision-making. As per participants, this includes the data owner (who is the data custodian), how the data will be used in the future, what decisions will be made, and how the data will be stored. An example of a quote includes.

The analysis in Table 3 revealed that 3 ECs, EC19, EC21 and EC2, with the agreement rates between 0.15 and 0.23, can be clustered under cross-platform integration and data synchronisation due to similarities in ensuring seamless operation of CCDT across several devices. These ECs mainly focused on on-demand data synchronisation for real-time or non-real-time data, capable of integrating with other devices such as a crowdsourcing mobile app to share information and communication between the physical and the virtual part, with a simultaneous data flow between them. These ECs were more prominently applicable to the consult category, with 03 identified ECs focused on enabling citizen consultation.

Data management and automation are the final theme identified based on the analysed ECs. The ECs identified within the theme were: use of hardware cameras, the use of hardware GPS, the integration of machine learning models allowing predictive analysis, comment management automating comment analysis and compliance with federal laws and workflow automation. Five of 15 ECs identified in Table 3 were considered under this theme, which focuses on providing opportunities for citizen consultation using CCDT. Use of hardware camera was the EC with the highest agreement rate (48%) among the five ECs identified under this category.

These five key themes were extended and integrated during the inductive coding process to provide an overall view of ECs for a CCDT and form a robust categorisation for ECs of a CCDT.

During the interviews mentioned in Section 3, participants were asked to identify additional ECs beyond those listed in the provided support materials (see Table 2). In the initial round, the participants identified 69 additional ECs. During the analysis process, the additional ECs were categorised and grouped based on the similarities or ideas presented by participants. An example of the process followed is explained.

ECs such as “Tagging and classification [P4]” and “Indexing of the comments and using artificial intelligence/machine learning to extract word tagging and classification [P3, P4]” were combined. The resulting EC on “Indexing of the comments and using artificial intelligence/machine learning to extract word tagging and classification” was considered in the final analysis. Furthermore, general comments that were made, such as advertising in social media [P20], in-person service [P20] and roles and responsibilities [P4], were considered strategies to promote the CCDT rather than ECs and were ignored. After examining the proposed ECs, the analysis finally derived 30 additional ECs spanning across several CRs of a CCDT platform.

Finally, the researchers carried out a comparative analysis by examining the implied arguments and justifications by all the participants against the initially emerging themes/ECs provided by the first participant using NVivo. As an example, P04 directly identified five additional ECs which were EC25, EC26, EC27, EC30 and EC33 (see Table 3). These ECs were later related to P08 based on the implied ideas provided during the interview and by examining the transcripts. Another example is that P18 pointed out the need for an error control/detection mechanism. While examining the transcripts of participants P2, P4, P8, P9, P14 and P19, it was understood that these participants had implied the necessity of error control and detection mechanisms. The process enabled us to carry out a robust analysis while ensuring the validity of the additional ECs proposed by the participants.

The emerging themes identified from the interviews are presented in Table 4 and discussed in Figure 2.

Despite the range of ECs suggested by the participants (Table 3), several arguments were raised in terms of understanding how ECs of a CCDT should be specific and structured while considering the common software requirements and DT maturity levels. The analysis of these data and the integration of findings in Table 3 led the researchers to derive a framework based on which the ECs of a CCDT should be structured. The framework comprises unique themes which diversify the framework developed by the current study to other digital twin use cases.

Figure 2 presents the proposed framework of this study. The key categories were identified by amalgamating the findings from the deductive coding process (Table 3) and the inductive coding process in Table 4. The initial layer of the framework emphasises the CRs identified for the CCDT. The next three tiers/layers filter the ECS, to be considered during the design of CCDT. Following the identification of CRs, ECs are identified in alignment with digital twin maturity models to enable the development of a digital twin. CCDT is developed for a wider community; hence, the necessity of considering the data privacy and security classification frameworks is vital to avoid revealing classified information to the public. This can impose constraints on meeting all the CRs identified by users of CCDT and aligned with Digital Twin maturity levels. The fourth layers contain five key categories within which the ECs that meet the data privacy and security classification standards and guidelines should be focused.

Comparison to the analysis presented in Tables 3 and 4, key themes emerged during the analysis of inductive codes presented by participants (Table 4). These included data privacy and security classification frameworks, accessibility and support services, resilience and fault tolerance and feedback management. This is mainly because while the literature contains ECs that are broadly considered within generic digital twins, a more specific CCDT requires ECs promoted towards ensuring collaboration and empowerment. To support the process, accessibility and support services, resilience, fault tolerance and feedback management are crucial. However, the execution of ECs to meet specific CRs should be guided by data privacy and data security classification frameworks to abide by state/country requirements. A detailed explanation of the emergence of the key categories is provided below.

This is the second layer of the framework. According to the participants, CCDT will be built upon the standard features as outlined in digital twin maturity levels. P20 stated that the definition of digital twins and their maturity levels will always include 3D data and real-time data through automated data collection methods like IoT devices, to then analyse to predict our future and forecast scenarios. A similar view was provided by participants P3 and P5 on 3D data and the need for physical and virtual connectivity in the digital twin environment.

The ECs identified in Table 4 incorporate IOT data sources, simulation and virtual representation capabilities and the ability to switch between 2D and 3D data and view photos over time (time-stamped photos) for predictive analysis or historical analysis were considered within this category. Some interesting thoughts were provided by participants on the ability of the digital twin to shift between 2D and 3D data.

Furthermore, the themes identified in Section 4.1 (Table 3) on the graphical user interface of the digital twin and cross-platform integration and data synchronisation could be considered within the maturity level of the digital twin. However, what is necessary to understand is how these ECs, which interrelate with the maturity level of digital twins, should be utilised against the requirements of citizen-centric capabilities. Five key themes to facilitate the process are discussed following the data privacy and security classification frameworks.

The third layer of the framework involves government information and classification, labelling and handling guidelines. These guidelines help developers and related parties assess the sensitivity and security of the information and manage labelling, handling, storing and disposing of the information correctly. This imposes constraints on the capacity of the CCDT to reveal information to the public, even if the CRs expect such capabilities. P20 stated that

It is vital to consider these guidelines before proceeding with the development of CCDT. Hence, the ECs suggested and identified during the interview under the key categories of enacting human trust on the platform, regulating human behaviour, accessibility and support services, Resilience and fault tolerance, and feedback management would have to be largely restricted within the requirements under the data classification framework.

Implementing standards and guidelines on security is one of the key ECs identified in Table 4. Furthermore, the ECs identified through deductive coding under the themes of preserving the security of the citizens and governance and data quality (Table 3) can be considered within this broader framework to enable the secured functionality of the digital twin. These features play a key role in ensuring human trust in the platform and are considered under the broader category of ECS to enact human trust on the platform. As per participant P18, trust could be interpreted as the ability of people being able to trust the digital twin:

P20 stated that it is essential to protect the citizens. As per the participant, the perception of considering digital twin technology and IoT devices to capture human behaviour in an environment is seen as a breach of privacy and using data against the citizen. Hence, human trust in the platform should be effectively catered to harness the true benefit of a CCDT.

The analysis process revealed that strict measures are required to avoid harmful behaviour in the CCDT environment. One potential concern the participants flagged was segregating CCDT from social media.

The inductive coding process (Table 4) revealed several ECs to regulate human behaviours in the CCDT platform. These include handling offensive and inappropriate comments and images through automatic detection and reporting features, automated scripts to identify abusive language, automated moderation or human interface were identified. The EC5 and EC22 were identified through deductive coding under the data management and automation theme (Table 3), highlighting the requirements of regulating human behaviour. Hence, these ECs were grouped to form a key category of ECs on regulating human behaviour.

The category of accessibility and support services emerged during the inductive coding process (Table 4) and 7 ECs proposed by the participants were largely categorised under accessibility and support services. Concerning the users of the platform being the general public with fewer technical specialities, ensuring accessibility and support service is crucial. Supporting video, written and audio help resources (eLearning, training Tooltips, in-context help and guidance), Accessibility considerations (Symbology, Cartology), progressive release of information (pop-up, tips) and in-context help and guide, AI-supported search queries, adding meta tags for virtual readers, different language entry possibilities and switch between audio and text or sub-titles were ECs identified within the category.

This mainly focused on the capability of CCDT to bounce back to the original condition following any abnormalities or errors in functioning. As per P19:

Architectural and infrastructure considerations for resilience and fault tolerance, metrics logging reporting and alerting, error control/error detection mechanism, backups/reroute to different data centres, IT infrastructure including powerful computing devices, reliable networks and high-speed Internet and processing power, minimum dependencies and ability to export or download selected pieces of data were considered as EC suggested by the participants, which was highly relevant within the theme. These ECs mainly emerge during the inductive coding process and are detailed in Table 4.

The capability of CCDT to seek feedback, thereby triggering public engagement and providing and managing feedback on different issues and concerns raised, is critical for the sustainable adoption of the system. During the inductive coding process, new ECs emerged that mainly fall into the group of feedback management. The ECs identified were systems to seek feedback and the ability to provide feedback in context within the web app or screen capture, indexing the comments and using AI and ML to extract word tagging and classification and communication policy or governance workflow around communication.

This study introduced a framework that allows researchers and developers to determine a CCDT based on user needs. The suggested framework is unique for a few reasons. Recent studies on digital twins-based user-centred designs have predominantly focused on organisational-level case studies. A few examples are Xiao and Zhang (2025) examine the adoption of digital twin technology in manufacturing by considering the roles of several stakeholders, including the government and customers. Tang et al. (2024) used QFD to convert technical requirements to design needs of a digital twin-based Prognostics and Health Management system. Newrzella et al. (2022) focus more on methodology to determine the most significant digital twin use cases with minimal work and great scalability. Himmelstob et al. (2023) used QFD has been used to design a digital twin for an asset administration shell for a component manufacturer. A CCDT includes diverse components (infrastructure and buildings), with citizens as the main contributors. Hence, it requires a robust approach towards user-centred designing that can integrate the heterogeneous components into a cohesive digital twin. Furthermore, the above-discussed studies have not explicitly identified the ECs that could deliver a user-centric output based on the CRs. As Liu et al. (2016) reveal, ECs for a digital twin have been extensively focused on characteristics such as physical objects, digital objects and the connection between the two. A CCDT aims to promote engagement, for which incorporating ECs based on CRs is required.

This study examined the application of ECs and developed a framework with key categories of ECs within which the CCDT should be devised to deliver a user-centric product. To do so, the study opted to integrate the concepts of QFD with spatially enabled digital twins of the built and natural environment and the spectrum of public participation to deliver specific results focused on CCDT. This integration is significant as it introduces an approach for designing the CCDT, with customers being the core source of determining the platform’s requirements. This resulting extended integration ensures that the CRs identified are relevant to citizen engagement, and the ECs lead to the development of a digital twin and finally create a CCDT. It has to be noted that poor requirements gathering, analysis and management account for 60–80% of the reasons why software projects fail (Bellisoy, 2019). Overcoming this drawback, our findings suggest ECs of a CCDT should be mainly framed within the key themes denoted in Figure 2 to allow citizens' requirements to be informed, consulted, involved, collaborated and empowered.

The framework in Figure 2 includes digital twin maturity levels as a key category on which the digital twin should be built, with standard features such as 3D modelling and real-time data (Aktürk and Irlayıcı Çakmak, 2024). However, the previous research on digital twin focus had been on deriving ECs with the ultimate goal of reaching digital twin maturity levels, which is one of the categories proposed by this study. In previous studies, Onan Demirel et al. (2021) developed a digital twin prototype of a cockpit and a fire emergency by incorporating ECs relevant to the use case. Similarly, Nie et al. (2021) focused on ECs for digital twin-based parking space management. Whereas Lim et al. (2020) considered ECs for a tower crane digital twin system. The above findings comply with the digital twin maturity levels proposed by the current framework. However, the above studies considerably vary in the context of other key categories of ECs in the proposed framework. These key categories, which diversify the proposed framework from other studies, are data privacy and security classification frameworks, enacting human trust on the platform, regulating human behaviour, accessibility and support services, resilience and fault tolerance and feedback management. Hence, this finding nuanced existing research by deferring a CCDT to other digital twins used at the organisation, functional and component levels, where digital twin maturity levels were mainly focused.

The suggested framework incorporates digital twin maturity levels. Visualisation of the digital twin models is considered the second level of the digital twin maturity model (ANZLIC, 2019), and it has 3D and 4D capabilities. This is evident in digital twin developments with user-centred design approaches, such as in Onan Demirel et al. (2021), Nie et al. (2021) and Lim et al. (2020). However, our findings suggest that in a CCDT environment, switching between 2D and 3D is vital in ensuring the platform’s usability. Therefore, unlike other digital twin use cases, a static model is not ideal in a CCDT environment. The findings suggest that 3D models might not be accessible to the general public as most people prefer to work and are familiar with 2D, with the capability to switch to 3D when required.

Data privacy and security classification frameworks are one of the top tiers to be considered after digital twin maturity levels in determining the feasibility of implementing CRs and relevant ECs (Figure 2). In a CCDT environment, public comments and photos play distinctive roles in maintaining the CCDT ecosystem. However, the capabilities of the CCDT would have to be limited despite the needs or expectations of the citizen as per state/country information classification, labelling and handling guidelines (Nsw Government, 2021). However, this is different from a facility-level digital twin (Nie et al., 2021), which largely relies on organisational principles and guidelines. This might challenge developers and researchers in meeting varying user requirements with due consideration of data guidelines. For example, in enabling CR 18, the risk of posting harmful comments or photographs and personal information could be considered high due to private disputes. This could lead to low-sensitive information reaching the sensitive information level under NSW guidelines. Hence, strict controls are mandatory to be aligned with state guidelines, and thereby, the CCDT capabilities should be limited within that boundary. This might pose a challenge to developers and researchers on how to meet varying user requirements with due consideration of data guidelines.

The proposed framework indicates the need for accessibility consideration for disabled individuals as another distinctive characteristic that is crucial in the CCDT environment. In a CCDT, users are the general public with diversified capabilities, contrary to other use cases, such as in a manufacturing environment, where the users would be technical staff with the necessary capabilities. In such an environment, considering the diverse user needs is vital. Han et al. (2019) proposed facilitating information on special requirements within the augmented reality application for tourism. However, the proposed framework emphasises the need to incorporate accessibility considerations to design the CCDT and provide equal opportunities for all users of the platform.

As per the digital twin maturity model presented by ANZLIC (2019), a CCDT could be distinct from other engagement applications with its capability to actuate bidirectional data and information flow and automate the process. However, as citizens are the core contributors to the platform, the need for ECs to govern and manage the inputs of citizens (comments, flags) cannot be overlooked. Therefore, the proposed framework has included regulating human behaviour and ensuring users' trust on the platform as a key category of ECs. Even though these ECs are not explicitly unique for digital twins in a CCDT, these ECs play a crucial role. Examples of such ECs include EC23, EC36 and EC40. With the recent increase in concern about how social media impacts children (US Senate Committee on the Judiciary, 2024), a CCDT needs to incorporate extensive measures and controls to manage citizens' input. Hence, unlike in other use cases, a CCDT in post-to-digital twin should incorporate these measures with utmost care.

A novel framework for determining the ECs of a CCDT was proposed by incorporating engagement capabilities to address a critical gap in the literature. The study used a rigorous methodology to collect data through semi-structured interviews and perform analysis using content analysis, thematic analysis and comparative analysis. The framework proposed by the study prioritises the user needs and suggests four main tiers based on which ECs for a CCDT should be identified. The five key categories identified at the fourth tier differentiate CCDT from other organisational or functional digital twins due to specific ECs considered to address the user requirements for CCDT at city scales. The five categories of ECs include enacting human trust on the platform, accessibility and support services, resilience, fault tolerance and feedback management. Existing research emphasises ECs to meet digital twin maturity levels, while the current framework further expands to consider other critical ECs based on the user needs of a CCDT. This is a key contribution to the evolving fields of digital twins in extended environments and geographical boundaries, as the framework focuses on specific design considerations to address the anticipated challenges in extended environments.

The novel framework introduced by the current study makes a significant contribution to theory by overcoming the limitations of existing studies that merely focus on informing citizens. This study derived ECs that enable users of a CCDT to be involved, consulted, collaborate and empowered. This addition to the theory provides a foundation for the core user expectations of a digital twin-based citizen engagement platform for designers and developers to consider. For example, the framework includes the considerations of data privacy and data security classification frameworks as one of the top-tier requirements within the framework. This ensures the ECs derived based on CRs are prioritised based on prevailing standards and frameworks to avoid complications during the development and post-development of the CCDTs. Also, the study’s theoretical contribution lies in integrating two principles to improve the QFD method to suit the user-centred design of digital twins. The first aspect relates to integrating principles for spatially enabled digital twins of the built and natural environment to frame the ECs and use the spectrum of public participation in devising CRs. This suggests that the decision-making process for user-centred design should be contextualised within the product’s specific application (such as digital twins or citizen engagement) rather than relying on conventional design methodologies.

The study provides practical contributions for the current digital twins available at city scales to incorporate citizen engagement as a core feature. Despite the current development in city digital twins, citizen engagement is overlooked due to challenges emerging from technology and managing citizen privacy and input. The ECs proposed by the study ensure the incorporation of citizen engagement capabilities while mitigating the common challenges. The study guides future developments for CCDT by introducing a framework for developers and other relevant parties to identify and prioritise ECs within the existing governance frameworks. Furthermore, the key themes of enacting human trust on the platform, regulating human behaviour, accessibility and support services, resilience and fault tolerance and feedback management suggested by the study provide key areas mandatory for CCDT-based developments to consider.

Further studies are recommended to assess the viability of the proposed framework in developing a CCDT platform or improving the abilities of the existing platform. Also, studies are required to assess the interdependence between ECs and CRs. This will support the prioritisation of the most crucial ECs based on the CR of the CCDT platform. A research endeavour would be to consider the impact of policies on the development of platforms such as CCDT. Existing policies and frameworks restrict the ability to gather data vital to the development of CCDT. Therefore, it would be valuable to address the limitations of policies and frameworks and explore technology-driven alternatives.

AI-based tools were not used in this study.

This research was approved by the University of New South Wales Human Research Ethics Committee on 10 May 2023 under the ethics approval number HC230186, and it was conducted following the ethical guidelines set forth by the Relevant Ethical Guidelines.

ANZLIC

The Australia and New Zealand Land Information Council

API

Application Processing Interface

CCDT

Citizen-Centric City Digital Twin

CR

Customer Requirements

EC

Engineering Characteristics

SPP

Spectrum of Public Participation

QFD

Quality Function Deployment