Comprehensive analysis of digital twins in smart cities: a 4200-paper bibliometric study

Digital Twin (DT) technology, a cornerstone of the Industry 4.0 era, represents a significant paradigm shift in how we interact with and understand physical systems and assets. Originating from Grieve's 2002 lecture at the University of Michigan (Grieves 2005) and later refined by NASA in 2010 (NASA 2010), the concept of DT has evolved into a sophisticated, multi-faceted approach to simulation and analysis. A Digital Twin is broadly defined as a digitally created virtual model of a physical object that leverages data to emulate the real-world behavior of the physical entity. It facilitates interaction and interoperability between the physical and virtual entities through interactive feedback, data integration, analysis, and iterative decision-making for optimized control, safety monitoring, and data analysis (Stark et al. 2017; Rosen et al. 2015).

Kritzinger et al. (Kritzinger et al. 2018) further categorized DT into three subtypes: digital model, digital shadow, and digital twin, each representing varying degrees of interaction and correlation between the physical and digital states.

The structure of a DT encompasses hardware and software components connected via middleware. The hardware typically includes IoT sensors and actuators, with the middleware playing a critical role in data management and communication between hardware and software. The software component, often an analytics engine, utilizes machine learning algorithms to transform raw data into actionable insights (Kritzinger et al. 2018). As depicted in Fig. 1, this system encompasses the various components constituting the digital twin architecture.

Digital twin system structure. This diagram illustrates the essential components of a Digital Twin system, showcasing hardware with IoT sensors, middleware for data management, and a software analytics engine

Full size image

Before going into the various applications of digital twins in various industries, it is important to comprehend the nature of digital twins and two closely connected ideas: digital model and digital shadow. These concepts are crucial for understanding this technology's depth (Kritzinger et al. 2018). The first concept is the Digital Model, a static digital representation of a physical object without automatic data exchange between the physical and digital entities. It can take many forms, including simulations, CAD files, 3D models, and mathematical algorithms. Digital models help design, optimize, and test by enabling the visualization, analysis, and manipulation of objects or systems in a digital context. A model is typically an estimation or prediction of how a system, process, or physical thing could function in the future or a certain setting. The second concept is Digital Shadow, which represents unidirectional information flow from the physical object to its digital counterpart, reflecting changes in the physical object. Through sensors, Internet of Things (IoT) devices, or other means, digital shadow gathers data from the asset (a database, a railroad system, or a banking platform). It delivers information that is fed into the model. This indicates that a digital shadow is current with the real object. It is helpful to understand it because it accurately depicts the asset in enough detail. The last concept is the Digital Twin, a dynamic, interactive digital representation capable of simulating, predicting, and interacting with data, showing a reciprocal impact between physical and digital states. Digital twins help with analysis, optimization, and predictive maintenance by simulating, monitoring, and controlling real-world systems or objects. They provide insights for enhancing effectiveness, dependability, and performance, as well as live feedback loops.

Figure 2 demonstrates the evolution from a basic digital model, lacking interactive data exchange, to a digital twin that dynamically mirrors and interacts with its physical counterpart, allowing for a two-way flow of information and continuous adaptation.

The progression from a static digital model to a dynamic digital twin, emphasizing reciprocal interaction between physical and digital assets

Full size image

Digital twins' ability to reproduce physical items, processes, and systems in a virtual environment makes them useful in various applications. This technology has applications across various sectors and domains, providing several benefits and chances for innovation. For example, in industry, they are used for predictive maintenance, optimizing energy usage in smart buildings, and simulating traffic patterns in smart cities.

In the IoT sector, DTs are pivotal, acting as a critical bridge between the physical and digital realms. They allow for the seamless integration of digital and physical entities, enhancing maintenance capabilities and improving equipment performance monitoring (Fang et al. 2022; Mihai et al. 2022; Hinchy et al. 2019; Guo 2020; Wang and Luo 2021; Rajesh et al. 2019; Revetria et al. 2019). IoT can be seen as the vehicle that drives data to Digital Twins, enabling these virtual entities to replicate and interact with their physical counterparts in real-time. Digital Twins depend heavily on IoT technologies for data acquisition. IoT devices like sensors, RFID tags, and smart wearables collect data from the physical environment, which the digital twin then utilizes for various analyses. This data integration facilitated by IoT is crucial for applications ranging from predictive maintenance in industrial settings to real-time monitoring and augmented reality applications (Rajesh et al. 2019; Revetria et al. 2019). As described in sources monitoring (Fang et al. 2022; Mihai et al. 2022; Hinchy et al. 2019; Guo et al. 2020; Wang and Luo 2021; Rajesh et al. 2019; Revetria et al. 2019), IoT's role is not just about data collection but also about ensuring seamless integration of physical and virtual worlds, thus forming the backbone of any DT system.

While the Internet of Things (IoT) plays a major role in shaping and augmenting the capabilities of digital twins, machine learning augments these capabilities by allowing digital twins to analyze data, forecast, identify anomalies, optimize performance, customize experiences, and learn and improve continuously. Integrating machine learning with DT technology enables real-time, autonomous analysis of extensive data streams, enhancing decision-making and optimizing asset and system performance (Rathore et al. 2021; Dong et al. 2019; Zohdi 2020; Jaensch et al. 2018; He et al. 2019). Machine Learning, a pivotal branch of Artificial Intelligence, involves algorithms that enable systems to learn and adapt from data without being explicitly programmed. Its relationship with Digital Twin technology is synergistic. Digital Twins, virtual replicas of physical entities, systems, or environments, require advanced analytical capabilities to process and interpret the vast amount of data they receive. This is where Machine Learning comes into play. Machine Learning algorithms in DT systems facilitate the autonomous, real-time analysis of extensive data streams. These algorithms are adept at detecting patterns, making predictions, and optimizing processes based on the data ingested from the physical assets that the digital twins mirror. For instance, Rathore et al. 2021 (Rathore 2021) highlighted how applying advanced AI techniques to data within a DT system enables the creation of an 'intelligent' digital twin. This intelligence is manifested in capabilities like predictive maintenance, operational optimization, and dynamic decision-making based on a continuous stream of sensor and virtual data. The application of various machine learning models, such as Deep Neural Networks (DNNs) or Genetic Algorithms (GAs), is contingent upon the specific requirements and use cases of the intended digital twins (Dong et al. 2019; Zohdi 2020; Jaensch 2018; He et al. 2019). Therefore, Machine Learning is not just a complementary technology for Digital Twins but a fundamental enabler of their advanced functionalities.

In the sector of smart cities, DTs are used for urban planning, traffic management, environmental monitoring, energy management, waste management, public safety, infrastructure maintenance, water management, healthcare, public services, tourism, citizen engagement, economic development, and climate resilience. They provide real-time data crucial for emergency response, optimizing public transportation, and ensuring efficient city operations (Allam and Jones 2021; Bouzguenda et al. 2019; Svítek et al. 2019; Yu et al. 2021; Ghosh et al. 2016). Smart Cities represent urban areas that integrate various electronic data collection sensors to manage assets, resources, and services efficiently. Digital Twins, within the context of Smart Cities, act as sophisticated tools for urban planning, management, and enhancement of living conditions. They utilize data gathered via IoT devices and analyze it using machine learning algorithms to optimize city operations and decision-making processes. Besides, they contribute to traffic management, environmental monitoring, energy distribution, public safety, and more (Allam and Jones 2021; Bouzguenda et al. 2019; Svítek et al. 2019; Yu et al. 2021; Ghosh et al. 2016). For example, digital twins utilize data from sensors and cameras to optimize traffic flow and public transportation systems in traffic and transportation management. They use real-time data to monitor air and water quality in environmental monitoring. In energy management, digital twins aid in the operation of smart grids and in identifying potential energy conservation areas. These applications underline the comprehensive impact that Digital Twins, empowered by IoT and ML, can have in transforming urban environments into more efficient, sustainable, and responsive entities.

The primary aim of this paper is to engage in a comprehensive bibliometric analysis, examining the evolving landscape of Digital Twin technology within Smart Cities. The study is dedicated to methodically examining the scholarly dialogue, identifying predominant trends, and revealing key themes and collaborative networks in this area. We aim to provide a detailed, structured understanding of Digital Twin technology's role in urban development, filling a notable void in existing literature reviews. The survey's distinctiveness stems from its thorough data-gathering approach for bibliometric analysis in the field of Digital Twin technology and Smart Cities, selecting the Web of Science database for its broad interdisciplinary coverage and meticulously filtering over 4,220 pertinent articles, enhancing the depth and scope of analysis in these domains.

Our research found that various publications in various literary works are advancing the DT idea. Because there are so many articles available, academics have also published several survey papers that aim to review the current state-of-the-art in digital transformation (DT) development, inform other innovators about potential research gaps, questions, and directions, and point the industry toward potential DT use cases that could yield substantial business value in their particular domain.

Current literature predominantly concentrates on applying digital twin technology within specific facets of smart cities. For instance, Jafari et al. (2023) and He et al. (2023) explore the utilization of digital twin (DT) technology in enhancing various sectors of energy management within smart cities, encompassing transportation systems, power grids, and microgrids. Weil (2023) delves into the infrastructure elements of digital twins in smart cities, focusing on storage, computation, and network components. Nica et al. (2023) investigates Multi-Sensor Fusion Technology's role in sustainable urban governance networks. Dani et al. (2023) introduces an architectural framework underpinning the flow for digital twin platform development aimed at urban condition monitoring. Lam et al. (2023) outlines a use case for the 3D visualization of a smart village in Busan, South Korea, employing a 3D Geospatial platform. Paripooranan et al. (2020) suggests augmented reality (AR)-assisted DT as a pioneering approach towards the future transformation of human-centric industries. Mora (2023) highlights the importance of incorporating innovation management theories into the exploration of smart city transitions, offering novel insights and practical approaches to enhance the governance of smart cities through an innovation management lens. Ariyachandra and Wedawatta (2023) provides an overview of digital twin technologies' implications on disaster risk management, addressing the challenges of implementing digital twins in smart cities. Additionally, several reviews, including those by Weil (2023) and Wang (2024), focus on bibliometric analyses concerning digital twins in the realm of smart cities.

This work aims to support the other existing survey initiatives and provide a comprehensive comprehension of the DT. The paper gives an in-depth overview of the DT idea, architecture, enabling technologies, applications in smart cities, challenges, performance metrics, datasets, software, and use cases for deploying DTs in diverse industries, complementing prior research. This paper aims to fill a critical gap in understanding the expansive and evolving field of Digital Twin technology and its integration into Smart City development. This study is driven by the need to systematically synthesize and analyze the burgeoning body of research in this interdisciplinary area, providing clarity and direction for future studies. This survey's uniqueness and unprecedented nature stem from its comprehensive and systematic bibliometric analysis of over 4,220 articles on Digital Twin technology and Smart Cities. A focused examination of specialized areas such as datasets, platforms, and performance metrics marks this distinctiveness. The rigorous methodology involving the Web of Science database ensures in-depth interdisciplinary coverage. The survey's meticulous approach in formulating search strategies and selective filtration of relevant articles contributes to its depth and breadth, making it a unique contribution to the field. The significant contributions of this survey paper are listed below:

• An overview of the DT definitions, concepts, and architecture in the literature

• A Detailed bibliometric study of over 4,220 publications in Digital Twin technology and Smart Cities, including thematic trends analysis like AI and IoT integration.

• Examination of datasets, platforms, and performance metrics specific to Digital Twins in urban settings and a critical evaluation of city models.

• Applications in Smart Cities: Exploration of Digital Twin technology applications in urban development, encompassing urban planning, energy management, and public health.

• Discussion of the challenges in implementing Digital Twin technology in Smart Cities, focusing on data integration, scalability, and security concerns.

• Outlining potential research avenues based on current findings, indicating areas for further exploration.

• Presentation of practical case studies demonstrating successful Digital Twin integration in urban development.

• Summarizing the main findings and implications and a call to action for further research in this evolving field.

The paper's organization follows a clear and structured approach, beginning with Section 1, an introduction that sets the stage for Digital Twin technology and Smart Cities. It progresses into Section 2, which provides a detailed bibliometric study, covering objectives, methodology, data collection, and analysis, leading to key findings and implications. Then, Section 3 explores the applications of Digital Twins in Smart Cities. Section 4 discusses some technological aspects of DT. Section 5 presents some examples of datasets and software for developing DT. Section 6 states digital twin performance metrics according to its structure. Section 7 addresses the challenges associated with digital twins. Section 8 introduces some case studies for DT. Section 9 discusses smart city governance in the era of digital twins. Finally, Section 10 summarizes the paper's conclusions and presents future research directions.

The primary objective of this research is to perform a bibliometric analysis (Yu and Merritt 2023) to acquire a comprehensive understanding of emerging topics, prominent journals, and evolving research trends associated with the application of digital twin technology in smart cities. Additionally, the study aims to shed light on the potential challenges and future research trajectories concerning digital twin technology in the context of smart city development.

2.1 Research methodology

This investigation employed a systematic literature review (SLR) to meticulously explore, assess, and integrate the extant body of knowledge regarding the designated theme, adhering to a rigorously defined protocol (Kyriazopoulou 2015). Adopting the SLR methodology is instrumental in delineating the contemporary scholarly landscape of a given topic, thereby uncovering existing research voids and delineating avenues for forthcoming scholarly inquiries (Kitchenham et al. 2009). The SLR framework comprises three pivotal phases: planning, execution, and dissemination. The research inquiries were articulated during the planning stage, and criteria for identifying pertinent literature and determining search strategies were established. The execution stage entailed the meticulous gathering and vetting of scholarly works in alignment with the previously established criteria. This phase was initiated with an initial screening of the collected records through their titles and abstracts to ascertain their pertinence to the posed research questions, followed by an in-depth examination of the full-text articles. A bespoke form was devised for the methodical extraction of data, capturing essential information from the chosen articles, such as facets of digital twin components, smart city innovations, and the research lacunae identified therein. Subsequently, the dissemination phase involved the analytical consolidation and synthesis of the compiled literature. The process was underscored by a commitment to transparency and precision, with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework guiding the data acquisition methodology (Liberati et al. 2009).

2.1.1 Research questions

To delineate the scope of the SLR, the following research questions guided the study:

• Q1: What are the components of digital twins in smart city applications?

• Q2: What are the existing technologies used in the development of smart city development based on digital twins?

• Q3: What are the research gaps and potential areas for future research?

2.1.2 Data collection

This section outlines the steps taken to collect relevant literature for the study. A PRISMA workflow diagram in Fig. 3 illustrates the study's search process. Initial literature searches were conducted in reputable databases such as Web of Science, Direct Science, and Scopus, which were chosen for their extensive coverage of scientific publications and advanced search capabilities. The research strategy applied an advanced search with keywords executed in the Web of Science and Scopus databases with a search string set to ("Digital twin," "virtual twin" or "virtual replica," and "smart city" or "smart cities"), for publications up to September 2023 and set to articles before 2018 were excluded. The period selected for the search is appropriate because there are few publications on digital twins and smart cities before 2018.

PRISMA workflow diagram

Full size image

2.1.3 Inclusion and exclusion criteria

In this survey, the inclusion and exclusion criteria were meticulously established to ensure a focused and relevant analysis in the fields of Digital Twin technology and Smart Cities. This careful selection was pivotal in delineating the scope of the study.

Inclusion criteria:

• Scope of content: Articles must focus on Digital Twin technology and its application within Smart Cities. This includes scholarly articles, conference proceedings, and review articles offering substantial insights into Digital Twin architectures, methodologies for Smart City implementation, technologies employed, and demonstrative case studies or laboratory setups.

• Language: Only articles published in English are considered to ensure the clarity and accessibility of the content for our analysis.

• Databases: Articles were sourced from the Web of Science, Scopus, and Direct Science databases to ensure a comprehensive and interdisciplinary coverage of the subject matter.

• Publication period: Articles published from 2018 to 2023 were included to capture Digital Twin technology's evolution and current state in Smart Cities.

• Detail requirements: Articles must present a detailed systematic architecture for a digital twin application and a clear methodology for Smart City implementation. They must also discuss the technologies used and provide demonstrative case studies or laboratory setups.

Exclusion criteria:

• Language limitation: Articles published in languages other than English were excluded to maintain consistency and comprehensibility in the analysis.

• Irrelevance: Publications unrelated to the direct intersection of Digital Twin technology and Smart Cities, lacking in detailed architecture, clear methodologies, technology discussion, or case studies, were excluded.

• Duplication: Duplicated records identified across the databases were removed to ensure the uniqueness and accuracy of the analysis.

• Date filter: Articles published before 2018 were excluded to focus the study on more recent developments and applications, reflecting the latest trends and innovations in the field.

The search produced 4,220 records from Web of Science, 382 from Scopus, and 24 from Direct Science. Through a preprocessing step, which involved removing duplicates and applying inclusion and exclusion criteria, 4,507 records were screened. This process yielded 4,073 articles that were deemed eligible for further analysis. The inclusion criteria were specifically targeted at articles that provided detailed systematic architectures for digital twin applications, methodologies for implementing smart cities, descriptions of technologies employed, and demonstrative case studies or laboratory setups.

2.2 Bibliometric study methodology

The methodology of a bibliometric study typically comprises a series of fundamental steps aimed at systematically analyzing scholarly literature within a specific field. These steps include formulating precise research questions to guide the analysis, identifying and selecting appropriate data sources, devising relevant search strategies using carefully chosen keywords, meticulously collecting and preparing the retrieved data, and employing established bibliometric techniques for rigorous data analysis. By adhering to this structured approach, researchers can effectively uncover trends and patterns in scientific publications and citations, thereby gaining valuable insights into the evolving landscape of their area of study (Mora et al. 2019). In line with these established practices, this research adopts a systematic approach for collecting, processing, and analyzing academic literature on digital twins within the context of smart cities.

2.3 Data analysis and visualization

This subsection outlines the methodologies and tools implemented to analyze and visualize the bibliometric data. For our study, VOSviewer was selected as the primary tool for managing and interpreting bibliographic data. We utilized network analysis methodologies to generate a range of visual representations. These included co-occurrence analyses, citation and co-citation maps, and keyword co-occurrence maps. Such visualizations were instrumental in uncovering patterns and discerning relationships within the collected dataset.

2.3.1 Publication trends

One of the key indicators in performance analysis is the annual number of publications. This metric serves as an indicator of research productivity. The data collected from 2011 to 2023 reveal a marked increase in publications focused on digital twin technology and smart cities. This surge in research output, demonstrating exponential growth, is depicted in Fig. 4. This Figure underscores the significance and escalating interest in this interdisciplinary area. Figure 4 illustrates the yearly publication rates concerning digital twins and smart cities. Additionally, Table 1 provides a concise statistical analysis of these findings.

The publications rate of digital twin and smart cities by year from 2011 to 2023

Full size image

2.3.2 Keyword analysis and research themes

This study's keyword co-occurrence analysis represents a systematic approach to understanding the prevailing keywords associated with digital twin technology and smart cities. The outcomes of this analysis, illustrated in Fig. 5, reveal a range of predominant research themes and technologies pertinent to the domain of digital twins and smart cities.

Visualization of keyword co-occurrence network

Full size image

Research themes in digital twin and smart cities:

• Theme 1: Integration of AI and big data analytics in digital twins

  This theme explores applying advanced deep learning techniques in processing and analyzing digital twin data. Key research areas are identified through terms such as "machine learning," "transfer learning," "simulation," "reinforcement learning," "cloud computing," "AI," "data analysis," "big data," and "forecasting."

• Theme 2: Integration of digital twins with IoT

  The focus here is IoT technologies, which are central to transmitting and collecting digital twin data. Relevant keywords include "wireless sensor network," "digital devices," "sensors," "5G", "communication," "wireless communications," and "monitoring."

• Theme 3: Energy management in digital twins

  This theme emphasizes the importance of energy efficiency and sustainability, highlighting keywords such as "energy efficiency," "energy utilization," "sustainability," and "renewable energy."

• Theme 4: Security concerns in digital twins

  Research in this area deals with the security aspects of digital twins, with keywords like "security," "privacy," "blockchain," and "fault diagnosis".

• Theme 5: Cloud computing and digital twins

  The final theme investigates the intersection of digital twins with cloud computing technologies, focusing on keywords such as "cloud computing," "edge computing," "fog computing," "blockchain," and "big data analytics."

Predominant technologies in the Digital Twin (DT) domain

Our study analyzed the authors' keywords to ascertain the most prominent technologies within the digital twin sphere. This investigation uses keyword frequency as a metric to identify the key technologies extensively employed in the digital twin (DT) domain. The term 'Internet of Things' (IoT) emerges as the most frequently cited keyword, demonstrated in Fig. 7. This finding underscores the pivotal role of IoT in the digital twin field, highlighting its extensive research coverage and the ongoing need for in-depth exploration of IoT applications to enhance digital twins' efficacy. Additionally, "AI" and "machine learning" are prominently used to analyze and process large volumes of digital twin data. Other notable technologies such as "cloud computing," "virtual reality," and "digital twin security" have also gained traction. Collectively, these technologies contribute to the efficient storage, visualization, modeling, and security of digital twin data. The data presented in the accompanying table and Fig. 6 substantiate the findings discussed in this subsection.

High-frequency keywords in digital twin research

Full size image

2.3.3 Analysis of geographical distribution

Examining the geographical distribution in the research and development of digital twin and smart cities technologies offers critical insights into the regional contributions, patterns of collaboration, and prospective areas for advancement. As depicted in Fig. 7(A), our analysis reveals a broad geographical spread in the field's research activities. We identified key regions contributing significantly to the field by utilizing a citation metric analysis on our dataset, which set a minimum of ten documents and fifty citations per country. China emerges as the leading contributor in terms of citations, followed by the USA, the UK, Italy, and Germany. Furthermore, Fig. 7(B) corroborates the leadership status of China, the USA, the UK, and Italy in this domain.

A Citations by country in digital twin and smart cities research. B Top 10 publishing countries in digital twin and smart cities research

Full size image

2.3.4 Analysis of source co-citation

The source co-citation analysis conducted in our study highlights the prominent sources within the domain of digital twins and smart cities. Of 49,275 sources, 433 met the established criterion of a minimum of 50 citations per source. The findings of this analysis are presented in Fig. 8. The most frequently co-cited journals include IEEE Access, the Journal of Manufacturing Systems, IEEE Transactions on Industrial Informatics, and IEEE Internet of Things. The analysis identified six distinct clusters, each represented by a unique color, as depicted in Fig. 7(A).

A Co-citation network of sources in digital twin and smart cities research. B Bibliographic coupling network among countries in digital twin and smart cities research

Full size image

2.3.5 Examination of international collaboration

The observed international collaboration in the digital twin and smart cities sector underscores research's global impact and relevance. Utilizing bibliographic coupling analysis on our dataset, with a set threshold of a minimum of 10 documents and 20 citations per country, 65 out of 103 countries met these criteria. A network visualization visually represents the bibliographic coupling among these countries in Fig. 8(B).

This analysis collated data on each country's publications, citations, and total link strength. Each node in the figure symbolizes a country whose size reflects its publication count. The visualization reveals that China leads in a collaborative network, boasting approximately 1151 documents and a total link strength of 871,379. Following China are the USA, the UK, and England. Notably, the USA's most extensive collaborations were with China, England, and India, while China's primary collaborations were with the USA, England, and Germany.

The colors in Fig. 8(B) delineate nine distinct clusters, indicating nations that frequently cite each other's research, suggesting closer collaboration within these groups. This mapping confirms that countries like China, the USA, the UK, Italy, and Germany are at the forefront in advancing research in digital twin and smart cities.

Digital twin technology offers a wide range of applications in smart city development, from optimizing traffic flow and energy usage to improving public safety, Environmental Monitoring and Management, Citizen-Centric Aspects, and Supply Chain Management and Enhancement. By creating virtual replicas of city infrastructure and systems, urban planners and policymakers can visualize potential changes and their impact before implementing them in the physical environment. Furthermore, digital twins can be instrumental in public safety by simulating emergency response scenarios and planning for effective evacuation routes in the event of natural disasters or other crises, as shown in Fig. 9. As the adoption of digital twins continues to grow, their role in shaping the future of smart cities will become increasingly prominent.

Applications of digital twin technology in smart city development

Full size image

3.1 Urban planning and management

Urban planning and management encompass the technical and political processes of utilizing land, infrastructure, and buildings within urban areas. This multifaceted domain includes urban design, land use, transportation, zoning, regulation, and environmental planning.

Urban planners and managers increasingly employ digital twin technology to enhance city functions like transportation and sustainability. Digital twins enable more informed decision-making and optimize planning, operations, finance, and strategy. In turn, such systems help reduce carbon emissions and expedite significant projects. Additionally, they enable the simulation of plans before implementation, allowing for the anticipation of potential challenges. The World Economic Forum 2022 recognized the role of digital twins in modeling future sustainable development by integrating digital technology with urban operational systems. This integration facilitates safer, more efficient urban activities. It creates low-carbon, sustainable environments through precise mapping, virtual-real integration, and intelligent feedback of physical and digital urban spaces (Yu and Merritt 2023).

Within urban planning and management, digital twins can represent entire cities or specific urban systems, assisting in various ways:

• Real-time Monitoring: Integrating sensors and IoT devices with digital twins provides real-time data on urban processes like traffic flow, energy consumption, and air quality.

• Simulation and Scenario Testing: Planners can use digital twins to simulate and test different scenarios, assessing the impacts of natural disasters or new transportation systems.

• Optimization: Analyzing data from digital twins can identify and address inefficiencies in urban systems.

• Public Engagement: Digital twins serve as interactive platforms for public involvement, allowing community members to view proposed changes and provide feedback.

• Maintenance and Asset Management: They enable tracking urban infrastructure conditions and predicting maintenance needs.

• System Integration: Digital twins facilitate understanding of interdependencies between various urban systems.

• Support for Decision-Making: Providing a comprehensive view of the city and its systems, digital twins enhance the decision-making process, ensuring decisions are informed by accurate, up-to-date information.

3.2 Energy management

Energy systems form the backbone of smart cities, ensuring the quality and functionality of these urban environments. This section delves into the application of DTs in energy systems, encompassing transportation systems, power grids, and microgrids.

Digital Twin technology finds varied applications in transportation systems. It supports transportation system infrastructures in several ways, such as monitoring transport systems, traffic forecasting, energy system management, predicting the energy consumption of electric vehicles, IoT-based parking management, analyzing driver behavior, forecasting subway regenerative braking energy, studying pedestrian behavior, controlling health systems, and detecting cyber-physical attacks. DTs can significantly contribute to these areas. For instance, using DTs for transportation system monitoring can reduce maintenance costs. Beyond modeling and planning, DTs facilitate optimal traffic management and provide accurate and extensive traffic and electric vehicle (EV) data, contributing to sustainable development and efficient urban traffic control.

In traffic management, DT technology has been utilized to predict patterns of energy consumption and production (Ketzler et al. 2020). Additionally, DTs play a role in IoT-based parking management, improving user services by saving time and reducing parking costs.

Various studies have employed DT technology in diverse contexts. In (Yan et al. 2022), authors analyzed real drivers' and pedestrians' behavior using DTs. In (Liu et al. 2020; Damjanovic-Behrendt 2018), DTs of drivers and vehicles were used in real-time to relay critical information to drivers and vehicles in the physical world. Moreover, in (Crespi et al. 2023), the Electric Vehicles (EVs) model employed DTs to monitor the behavior and optimally manage charging programs, using energy consumption parameters and charging capacity and frequency for modeling the virtual twin.

A microgrid is an autonomous energy system characterized by distributed energy resources and interconnected loads. It functions as a manageable entity within the larger grid, enabling it to operate in either island mode or in conjunction with the grid (Ton and Smith 2012). The objective of microgrids is to enhance the functionality of energy systems in terms of sustainability, economic viability, efficiency, security, and overall energy management. Key aspects of microgrid performance include reliability, self-sufficiency, security, flexibility, and optimality. Studies on microgrids utilizing the Digital Twin (DT) framework have encompassed areas such as forecasting (Din and Marnerides 2017; He et al. 2017), management and monitoring (Xu et al. 2019; Park et al. 2020), fault prediction (Nowocin 2017; Goia et al. 2022), and security (Huang et al. 2021).

The development and implementation of DT-based power grids are instrumental in improving network behavior under various conditions. Network studies employing DT include diverse analyses such as restoration (Biagini et al. 2020), reliability (Podvalny and Vasiljev 2021), prediction (Park et al. 2020), addressing uncertainty (Raqeeb et al. 2022), energy hub management (Kuber et al. 2022), and ensuring both physical and cyber security. Each of these analyses offers unique insights into network behavior. In reference (Endsley 2016), Situation Awareness (SA) is the ability to perceive the elements in a specific environment, understand their properties, and anticipate their future statuses. SA is crucial in augmenting decision-making, especially in complex systems like the Energy Internet of Things (EIoT) (He et al. 2023). It provides essential information critical for such systems' operation, enhancing efficiency and effectiveness.

In the application of Digital Twin technology in power systems, several significant challenges emerge:

• IT infrastructure limitations: Existing infrastructure often falls short in supporting the data analysis demands of DT environments.

• High-performance computing needs: Utilizing high-performance GPUs and cloud services from major providers is essential for adequate support.

• Connectivity issues: Software errors and power outages present obstacles in real-time monitoring.

• Cybersecurity risks: The extensive data exchange in DT systems heightens vulnerability to cyber-attacks, necessitating secure platforms.

• Standardization requirement: The absence of standardized protocols impedes DT development, highlighting the need for unified approaches for model definition, storage, and execution.

The exploration of digital twin applications in energy management reveals several key areas for future development:

• Advancements in big models: Addressing challenges in AI, such as limited model generalization and the need for high-quality data, by developing larger, more adaptable models.

• Virtual twin structures in power systems: Detailed modeling of power system entities using virtual twins, enabling dynamic visualization and strategy development for urban transformation.

• Application of theoretical models: Utilizing chaos and complex system theories (Mir et al. 2022) to understand and optimize the nonlinearities in power systems, offering a novel approach to managing system complexities.

3.3 Traffic and mobility management

Traffic and Mobility Management in Smart Cities (Xu et al. 2023), enhanced by Digital Twin (DT) technology, represents a significant advancement in urban planning and logistics. DTs enable:

• Real-time traffic simulation: Mimicking urban traffic flow to identify and alleviate congestion points.

• Public transportation optimization: Analyzing patterns to improve transit routes and schedules.

• Pedestrian flow management: Ensuring safer and more efficient pedestrian movement.

• Pollution reduction: Aiding in strategies to lower emissions through traffic regulation.

• Emergency response enhancement: Assisting in quicker and more efficient routing for emergency services.

• Data-driven decision making: Utilizing sensor data for informed traffic management decisions.

• Sustainable urban planning: Contributing to long-term urban sustainability goals through efficient mobility solutions.

These applications of DT in traffic and mobility management significantly contribute to creating more livable, efficient, and sustainable urban environments.

3.4 Environmental monitoring and management

Digital twin technology is increasingly integral in urban development, offering real-time insights and solutions for environmental management:

• Optimization and prediction: Digital twins, as virtual representations of physical entities, enable process optimization, change monitoring, and future scenario prediction (Wang et al. 2023).

• Environmental monitoring applications: Usage in water quality monitoring, detecting pollutants, and adapting to changing environmental conditions.

• Data integration in smart cities: Interconnection of multiple digital twins, using diverse data sources like temperature and humidity, to forecast environmental conditions (Ivanov et al. 2020).

• Sensor utilization: Various sensors capture essential environmental data for digital twin construction, including Kinect v2 depth cameras and electronic gloves for manufacturing systems (Nikolakis et al. 2018).

• Food industry monitoring: Application in monitoring and predicting food quality, employing wireless sensors for environmental factors like humidity and temperature (Defraeye et al. 2019).

• Agricultural management: Use in agriculture for crop growth monitoring and simulating interventions, aiding in remote farm management (Verdouw et al. 2021).

• Healthcare applications: Implementation in healthcare for environmental monitoring and mental health management using smartwatch sensors (Bagaria et al. 2019).

These diverse applications showcase the role of digital twins in enhancing urban planning, agriculture, healthcare, and more.

3.5 Public health and safety

In the development of Smart Cities, Digital Twin Technology plays a crucial role in enhancing public health and safety (Erol et al. 2020). It offers a dynamic and integrated approach to managing complex urban health challenges through simulation and analysis. This technology's applications include:

• Disease outbreak prediction and management: Leveraging real-time data to simulate disease spread and plan responses.

• Emergency preparedness: Using simulations for natural disasters or public safety incidents to enhance response strategies.

• Resource optimization in healthcare: Improving the allocation of healthcare resources like hospital beds and emergency services.

• Environmental health monitoring: Tracking and analyzing environmental factors that impact public health, such as pollution levels.

• Public safety and incident response: Simulating various scenarios to optimize law enforcement and emergency services.

These applications demonstrate the transformative impact of Digital Twin Technology in public health and safety within Smart Cities.

3.6 Citizen-centric aspects

Technological advancements have focused on urban development and infrastructure management, employing physical sensors such as the Internet of Things (IoT) and satellites (Borrmann et al. 2018). However, not all city digital twin implementations have citizen engagement in mind. A citizen-centric digital twin (CCDT) approach views citizens as integral components of a data-driven city, with human sensors playing a key role in addressing city-scale challenges (Saeed et al. 2022). This approach distinguishes itself from traditional digital twin frameworks by prioritizing citizens as the central element and integrating technologies like processing, data acquisition, and visualization to enhance citizen involvement in infrastructure governance.

Developing a CCDT requires the execution of numerous processes and technologies. One such technology involves sensors like Volunteered Geographic Information (VGI) (White et al. 2021), which transfer data from the actual city to its digital twin, followed by analysis using various analytical tools. Managing data from diverse sources at a city scale presents challenges in scalability, reliability, and the performance of real-time analytics and modeling (Langenheim et al. 2022).

The study by Abdeen et al. (Abdeen et al. 2023) indicates a scarcity of publications in this field over the past five years. However, a rising trend in CDT interest post-2017 was noted, with publications doubling by the end of 2022 compared to 2019. Research works (Ford and Wolf 2020; Fan et al. 2021) discuss the application of digital technologies in catastrophic situations and emergency responses. The capabilities of intelligent digital twins in various application fields have been examined (Shahat 2021; Deren et al. 2021), with the latter focusing on hazards like epidemic services, traffic control, and flood monitoring. (Shahat et al. 2021) concentrated on data simulations, fusion, administration, and collaboration. (Charitonidou 2022) addressed citizen participation in decision-making, highlighting that limited variables and processes and overlooking social aspects of urban contexts can render citizen input integration ineffective.

In the literature, various data acquisition mechanisms are employed to support CCDTs. One prominent method is the use of open-source data platforms (OSDP), providing spatiotemporal performance data relevant to CCDT applications like disaster management (Ghaith et al. 2022) and public services monitoring (Diakite et al. 2022). However, the effectiveness of CCDTs can be compromised if data is unreliable. Another mechanism is crowdsourcing, which generates large quantities of data and is particularly useful when remote or IoT sensors are unavailable (Trusov and Limonova 2020). Nevertheless, citizens' data errors or bogus inputs can affect CCDT effectiveness (Trusov and Limonova 2020). Visionary concepts for disaster city digital twins with extensive data (images, text, geo maps) have been proposed to enrich CCDT content (Fan et al. 2021).

Remote sensors are effective in modeling 3D city aspects of CCDTs and for large-scale urban monitoring (Fan and Mostafavi 2019; Fan et al. 2020). Geospatial platforms storing and managing data from individual vehicles or pedestrians have been proposed (Lee et al. 2022), though data accessibility to all stakeholders remains challenging for CCDT integration. Furthermore, IoT sensors (Nochta et al. 2020), deployed in large numbers and integrated effectively, facilitate urban data monitoring but require advanced communication infrastructure.

Advanced AI algorithms also play a crucial role in CCDTs, enhancing citizen engagement. Genetic algorithms (Fan et al. 2021) have been used to study the range of disruptions during hazard events, while Convolutional Neural Networks (CNN) (Pang et al. 2021) and Burst detection algorithms (Fan et al. 2021) help analyze crowdsourced data and social media frequencies providing insights into citizen perspectives and infrastructure governance through CCDTs.

3.7 Supply chain management and enhancement

The supply chain, encompassing the entire spectrum from raw material sourcing to the distribution of finished products, has seen a transformative integration of Digital Twin technology in recent years. Digital Twins, as virtual models of physical assets, offer real-time monitoring, analysis, and optimization across all facets of the supply chain, ranging from procurement to distribution (Tao et al. 2017). According to van der Valk et al. (van der Valk et al. 2022), these digital replicas enable two-way data exchange between the digital and physical worlds, providing professionals with exceptional visibility and traceability. This level of insight facilitates the identification of complex behavioral patterns and proactive problem detection, which is crucial for maintaining operational continuity.

As Gerlach et al. (Gerlach et al. 2021) highlight, Digital Twins are instrumental in offering real-time inventory insights, enabling the simulation of various scenarios, and assisting in planning and forecasting. These capabilities can result in significant cost reductions and process efficiency improvements. The study by Srai et al. (Srai and Settanni 2019) explores the optimization opportunities that Digital Twins offer in areas such as transportation resource management, demand–supply analysis, customer service improvement, and revenue enhancement. They also emphasize the role of technology in identifying and addressing inefficiencies.

The influence of Digital Twins in improving stock availability, a key aspect of manufacturing operations, is underscored by Abouzid et al. (Abouzid and Saidi 2023). Furthermore, (Lugaresi et al. 2023) introduces the concept of "technological labelers" like IoT devices, cloud computing, and advanced analytics, which are crucial in developing a comprehensive digital twin of a company's value chain. IoT integration, in particular, is noted for significantly enhancing supply chain efficiency by providing real-time data and contextual insights. In addition, distorted demand signals can result in various supply chain challenges, which can be effectively addressed using Digital Twin technology (Abouzid and Saidi 2023).

In conclusion, the discourse delves into the exploration and development of digital twins within automated manufacturing systems, showcasing the expansive potential of this technology in modernizing and streamlining supply chain management processes (Abideen et al. 2021).

This section explores the fundamental components and emerging advancements in digital twin technology, covering various technological aspects essential for understanding Digital Twins across disciplines. Each component offers unique insights into critical subjects, including distinctions between Digital Twins and Building Information Modeling (BIM) and the development of Cognitive Twins. By studying frameworks like the Five-Layer Architecture and advancements such as Cloud and Edge Computing integration, this section aims to reveal the technological foundations driving the evolution of Digital Twins. Readers will gain a deeper understanding of the technological breakthroughs shaping the future of Digital Twins and their applications through this comprehensive examination.

4.1 Distinction between digital twins and Building Information Modeling (BIM)

In construction technology, the emerging potential of digital twins and the rapid advancement of smart technologies has garnered significant interest. Although 'digital twin' is a relatively new term in the construction research literature, it is often conflated with Building Information Modeling (BIM), leading to some conceptual ambiguity. It is imperative to clarify the differences between these two concepts.

Building Information Modeling (BIM) is a digital representation of a building or structure's physical and functional characteristics. It is a tool architects, engineers, and construction professionals utilize to create detailed digital models of buildings, encompassing all systems and components, including architectural, plumbing, electrical, and HVAC systems. BIM enables the creation of accurate construction plans, virtual walkthroughs, and performance testing under various scenarios. (Wang and Meng 2021) defined BIM as a method that integrates geometric and non-geometric data. The 3D model, often called the BIM model and realized through object-oriented software, is a critical component of BIM (Cerovsek 2011). However, BIM primarily manages static data and requires external technologies to update models with real-time data (White 2021). In construction projects and asset management, a vast amount of non-geometric data is essential for informed decision-making but is often underutilized (Khudhair et al. 2021). BIM models have limited capacity to handle large volumes of dynamic and multifaceted data, necessitating advanced storage and processing technologies. These limitations can lead to data underutilization, inefficient decision-making, and financial implications. The advent of digital twin technology offers a solution to overcome these constraints inherent in BIM.

Digital twins and BIM represent two distinct technological applications in the construction sector, differentiated by their functions. BIM is most effective in the design and construction phases, while digital twins excel in building maintenance and operations. A digital twin system involves data linkage that transfers information between the physical asset and its virtual counterpart. This indicates that a BIM (Building Information Modeling) model is the initial step toward developing a digital twin in the construction industry. Digital twin technology integrates the BIM model with the physical world, enabling bidirectional data exchange. This connection allows for the real-time updating of the BIM model, enhancing asset implementation and management decision-making. The synergy between BIM and Digital Twin technology has the potential to revolutionize the construction industry. By combining the detailed architectural and structural information provided by BIM with the real-time operational data and analysis capabilities of Digital Twin technology, construction professionals can create comprehensive, accurate digital models of buildings. These models can then continuously monitor and optimize building performance in real-time.

4.2 Framework of the five-layer architecture in digital twins

A digital twin, essentially a digital representation of a physical entity, process, or person contextualized in a virtual environment, is a pivotal tool for organizations to simulate real-world scenarios and outcomes, thus enhancing decision-making capabilities (Moosavi et al. 2021). As shown in Fig. 10, the architecture of digital twins is typically structured into five principal layers, as outlined in (Jones et al. 2020):

Five-layer architecture in digital twins

Full size image

1. 1.

   Physical layer: This foundational layer comprises the actual physical objects or entities. It utilizes sensor technology for data acquisition and can receive commands from the virtual layer. This layer provides real-time data feedback to the digital twin model.

2. 2.

   Data sensing layer: Responsible for collecting diverse information types, this layer employs various sensors to monitor the system's status and operational process in detail. The data heterogeneity and variety stem from diverse data generation sources, such as IoT sensors, information systems, and wearable devices.

3. 3.

   Data transmission layer: As a crucial link, this layer ensures data transmission between the physical and virtual layers. It leverages communication integration protocols and interactive security technologies to facilitate this transfer (Lohtander et al. 2018).

4. 4.

   Virtual layer: In this layer, the components of the real world are digitally reconstructed. It builds a collection of digital twins using data transmitted from the physical layer, enhanced with historical or integrated network data. This layer dynamically tunes itself based on the real-time data from the physical layer and can be influenced by modifications made in the application layer.

5. 5.

   Application layer: This layer visualizes the data and simulations derived from the virtual layer, presenting a graphical model that staff can easily interpret. Modifications in the physical or virtual layer parameters can lead to simulation changes. These can then be revised and optimized based on the observed or extrapolated results.

Each layer in this five-tier architecture plays a distinct yet interconnected role, collectively enabling the digital twin to function as a comprehensive, dynamic system for analyzing, simulating, and enhancing real-world processes and entities.

4.3 Integration of cloud and edge computing in digital twin environments

Cloud computing represents a large-scale computational approach that leverages the Internet to facilitate sharing computing, storage, and other resources, accessible anytime and anywhere on demand. In contrast, edge computing is a novel computational model that processes a portion of data using distributed computing, storage, and network resources between data sources and cloud computing centers.

Edge computing is increasingly recognized for its potential to enhance privacy, reduce latency, conserve energy and costs, and boost reliability. It is particularly well-suited for Digital Twin (DT) scenarios that demand low latency, high bandwidth, high reliability, and stringent privacy measures. In DT-assisted edge computing setups, the framework includes user devices, edge servers, resource devices, and the DT itself. User equipment initiates task requests to the edge server, which then allocates computing resources to the task, with the DT deployed within the edge server. Reviewed literature demonstrates the application of Cloud and Edge technologies in various contexts. Cloud storage is universally employed in these studies. Earlier research also utilized cloud computing for user interfaces, with cloud-rendered 3D models, as indicated in (Xu et al. 2021), or through GUIs accessible via web applications as in (Urbina Coronado 2018). While initial studies relegated intensive data processing and analytics to cloud computing due to its superior resource access, recent advancements in edge device capabilities have led to the implementation of edge computing, employing techniques for heavy data analysis or machine learning (Cathey et al. 2021; Lu et al. 2021a; Zhang et al. 2022).

Recent digital twin research, such as (Alam and El Saddik 2017), employs edge computing, describing a framework where each device is represented as a cloud-based digital twin. This hierarchical architecture involves higher-level digital twins composed of simpler units in a master/slave relationship, enhancing the communicability of traditional cyber-physical systems with cloud servers' advanced computational and storage capacities. Focus on edge-based architectures is evident in (Dong 2019; Lu et al. 2021a), with research by Dong (Dong 2019) on enhancing energy efficiency in 5G services through deep neural networks and Lu (Lu et al. 2020) exploring the use of digital twins in-network replication and machine learning via federated learning.

In industry, studies such as (Lu et al. 2021b; Zhang et al. 2022) concentrate on Smart Vehicles, driven by the rise in edge computing power. Other research, including (Liu et al. 2019; Martinez-Velazquez et al. 2019), investigates the application of digital twins in healthcare, aiming to provide high-quality, real-time care to senior citizens. Most other studies, such as (Xu et al. 2021; Urbina Coronado et al. 2018; Hu et al. 2018; Bellavista et al. 2021), are categorized under Smart Manufacturing, focusing on industrial productivity improvements. Cloud-based digital twins play a crucial role in optimizing IoT device energy consumption and operational efficiency (Li et al. 2020), detecting and preventing potential system failures (Cathey et al. 2021), and ensuring data privacy and integrity (Wen et al. 2020). Thus, cloud computing and IoT emerge as complementary technologies, synergistically advancing the development of smart, interconnected systems.

Research in the Oil and Gas industry reflects a systematic adoption of digital twins and cloud/edge computing. For instance, (Pivano et al. 2019) discusses offloading simulations and data analysis to public cloud servers to access greater computational resources and avoid complex local IT infrastructures. Tygesen et al. (Tygesen et al. 2018) highlight the role of high-performance cloud computing in wave load modeling, which is essential for maintaining offshore platform integrity. (ASME 2018) describes the use of cloud data lakes for data verification and physical model feeding. At the same time, a microservices-based approach has been presented for designing and implementing digital twins using open-source tools (Zborowski 2018).

4.4 Implementation of augmented reality in digital twin technology

Augmented Reality (AR) is a technology that merges the real with the virtual, facilitating real-time interaction and 3D registration (Damjanovic-Behrendt and Behrendt 2019). It enhances user experience by superimposing graphics, video streams, or holograms onto the physical world (Yin et al. 2023). It is supported by various devices such as AR head-mounted displays (HMD), tablets, head-up displays (HUD), projectors, VR HMD with cameras, and 2D screen augmentations.

AR's enhancements are primarily derived from its visualization, interaction, 3D registration, and information collection capabilities as a unified device (Billinghurst et al. 2015). AR contributes to Digital Twin (DT) technology in several dimensions. In the virtual twin dimension, AR provides visualization of non-registered geometry, data, workflows, and basic status monitoring and alerting for operators. It also allows users to update DT information, for example, by scanning barcodes or adding annotations. However, the full potential of AR in augmenting DTs remains underexploited. In the hybrid twin dimension, AR enables multi-modal interactions and on-site registered visualization, with a need for further exploration and utilization in cyber-physical interaction functions. In the cognitive twin dimension, AR-assisted DT, bolstered by edge-cloud computing systems, is poised to play a more significant role in areas like visual programming, human–robot collaboration (HRC), product design, and human ergonomics, marking promising future directions for AR-assisted DT.

Applications of AR-assisted DT span a wide range of physical scenarios, encompassing the entire product lifecycle, including the management of production facilities and services. The production process and service phases include design, production, distribution, maintenance, and end-of-life stages, as illustrated in Fig. 11.

Applications of AR-Assisted digital twins in engineering lifecycle, including design, production, distribution, maintenance, and end-of-life stages

Full size image

The systematic design process involves prototyping, pilot runs, and testing operations. Real-time data from product usage, collected by sensors, informs smart product service design or redesign, integrating DT in mapping virtual and physical objects (Praschl and Krauss 2022). Research in AR device utilization for design falls into three categories: product design (Zheng et al. 2018), service design, and system design. Service design, a creative and user-centric process for enhancing or creating new services (Chang et al. 2020), is supported by several studies focusing on operation training (Blomkvist et al. 2023), driving and flight guidance (Moya et al 2020b), smart environments (Vidal-Balea et al. 2021), smart cities (Lacoche and Villain 2022; Ssin et al. 2021a), and smart wetlands (Ssin et al. 2021b), aiming to deliver user-centered services that cater to the needs of users and stakeholders.

System design entails developing architecture, components, and core algorithms for AR-assisted DT scenarios. Moya et al. (Aheleroff et al. 2020; Moya et al. 2020a) introduced two self-learning DT systems with screen augmentation for fluid behavior prediction and beam load analysis.

The production process includes goods fabrication or service provision, subdivided into process planning and scheduling (Wiegand et al. 2018), monitoring and control (Lemos et al. 2022), assembly (Kritzler et al. 2017), and robotics-related works (Židek et al. 2021). Real-time machine status monitoring and interactive control are prevalent in research, as demonstrated by Paripooranan et al. (Paripooranan et al. 2020), who developed an AR-enabled 3D printer DT for alerting abnormal statuses.

In distribution, warehouse management utilizes AR and DT, as shown by Petković et al. (Petković et al. 2019) in their use of a warehouse system DT (comprising the warehouse, automated guided vehicles (AGV), and operators with AR HMD) to test a human intention estimation algorithm.

Maintenance work encompasses various strategies and can be categorized into reactive, preventive, and predictive maintenance (Petković et al. 2019), adopting different approaches within the AR-assisted DT framework.

4.5 Hybrid twins in mixed reality applications

Mixed Reality (MR) applications offer an interactive experience that blends real and virtual environments, akin to Augmented Reality (AR) (Damjanovic-Behrendt and Behrendt 2019). Additionally, the term Extended Reality (XR) encompasses Virtual Reality (VR), AR, and MR and has been included in the research scope. The enhancements brought about by AR are examined across three distinct dimensions of the digital twin: the virtual twin, hybrid twin, and cognitive twin, as depicted in Fig. 12.

Layered framework for digital twin classification based on augmented reality devices' perceptual capabilities

Full size image

The virtual twin dimension encompasses data transmission from physical to virtual realms, non-registered visualization, and essential status monitoring and alerting functions based on sensor data. When enhanced by Augmented Reality (AR) devices' perceptual capabilities, this dimension can improve the data transmission process from the physical to the virtual space and suitably update Digital Twin (DT) information. Beyond IoT sensor data, on-site information such as barcodes and workspace details can also be gathered through AR applications, exemplified in warehouse management (Xia et al. 2022).

Reference (John Samuel et al. 2022) discusses the concept of hybridization in DTs, focusing on refining DT accuracy through self-adaptation and data-driven estimation techniques. This approach integrates physics-based model predictions with process measurements, creating a hybrid digital twin (HT) that facilitates the soft-sensing of otherwise hard-to-predict data.

The hybrid twin dimension emphasizes analysis and feedback from the virtual to the physical world, such as context information-related analysis, visual registration, multi-modal interaction and control, and the functionalities based on these aspects. Traditional DTs manage real-time data analysis, including simulation, prediction, diagnosis, and optimization, feeding back the analysis outcomes from the virtual to the physical world. AR-assisted DTs enhance this analysis with on-site data, adding capabilities like object localization, scene understanding, and cyber-physical interaction computation. For instance, in human–robot collaboration (HRC) assembly (Johansen et al. 2023), the hybrid twin dimension offers immersive visual registration beyond traditional 2D interfaces, displaying geometry and key data overlaid on the physical entity in the correct position. In contexts such as assembly (Liu et al. 2022; Zhao and Sun 2020), maintenance (Meier et al. 2021; Li et al. 2021; Rabah et al. 2018), and manual or semi-automated tasks (Koteleva et al. 2021; Rebmann et al. 2020; Mandl et al. 2017), operators can reference on-site instructions and guidance to work more efficiently. Additionally, geometry overlay for inspection (Catalano et al. 2022; Xie et al. 2020) or motion preview (Kim and Olsen 2021) aids operators in verifying the shape or movement of physical entities against planned outcomes. Users can also add geometry-linked or position-related annotations through AR.

Akroyd et al. (Akroyd et al. 2022) introduced the concept of the Universal Digital Twin, a digital twin that leverages a dynamic knowledge graph to enable cross-domain interoperability for DTs.

4.6 Development of cognitive twins in digital twin technology

Cognitive twins represent an advanced form of digital twins endowed with high-level cognitive capabilities encompassing machine and human intelligence. These cognitive twins are designed to address complex and unpredictable situations using enhanced computational power dynamically. Augmented Reality (AR) significantly contributes to the development of cognitive twins as it can function as a wearable computational unit within the edge-cloud architecture (Li et al 2022). HoloLens 2, a widely-used AR device, notably possesses substantial computing power (1 T FLOP) compared to wearable devices like sensors. This capability allows training models on high-power devices and their subsequent deployment on HoloLens 2, highlighting one of AR's key benefits to digital twins.

Cognitive Digital Twins (CDTs), originating from the domains of Industry 4.0 and Smart Cities, are recognized for their ability to support autonomous activities (Um et al. 2018; Liu et al. 2023; Zheng et al. 2021). Semantic technologies, including ontology and Knowledge Graph (KG), are vital in interlinking digital twins in virtual spaces. These technologies eliminate ambiguity across heterogeneous systems, thus enhancing digital interoperability and enabling cooperative decision-making (Rožanec et al. 2022). As defined by (Pan et al. 2021), ontology involves a set of formal and explicit vocabularies characterized by shareability and reusability, describing domain-specific knowledge, entities' attributes, and their interrelationships. While early research primarily focused on utilizing ontology for data modeling and sharing (Rožanec et al. 2022), recent studies emphasize that integrating semantics with digital twin technologies can advance the capability and interoperability of CDTs in autonomous and cooperative decision-making (Zheng et al. 2021).

The knowledge graph has become increasingly important in developing and managing CDTs because it can delineate relationships between real-world entities or link data (Liu et al. 2023). For instance, recent research has explored using knowledge graphs and digital twins in managing assets and tasks in smart manufacturing systems (Guarino et al. 2009) and underwater ship inspections (Zheng et al. 2023). Some studies have concentrated on methodologies that leverage knowledge graphs to create semantic data models for shaping digital twins (Waszak et al. 2022).

Furthermore, the evolving flexibility and customization in futuristic smart manufacturing are closely linked with human intelligence. For example, in human–robot collaboration (HRC) tasks aimed at improving human ergonomics (Steinmetz et al. 2022), operators can adjust robot poses through gesture-based interactions with the robot's digital twin. After receiving instructions from human operators, the robot digital twin learns to perform better and meet human needs. Additionally, in the timber prefabrication process (Dimitropoulos et al. 2021), AR provides effective interaction methods to enhance mutual understanding between operators and collaborative robots, ultimately facilitating harmonious task sharing.

4.7 Classification of digital twins by scale

Digital twins can be categorized into various types based on their scale and comprehensiveness, including component, asset, system, and process twins (Amtsberg et al. 2021).

1. 1)

   Component twins: This approach suits large, complex digital twins. The adaptation and uncertainty quantification of the model in such applications can be framed as a Bayesian state estimation problem. Here, data from the physical world is used to infer which models from a model library best represent the digital twins. This approach strategically selects specific components for replication in the digital twins to avoid data redundancy and reduce costs. Microsoft has developed the Azure Digital Twins (ADT) platform (Cinar et al. 2020), facilitating model creation and offering a graph API for querying and interacting with these digital twins. The ADT platform enables users to visualize and examine the relationships among components, such as creating 3-D digital twins of a factory with a user-friendly interface. This interface allows operators to monitor the state of each machine. A notable challenge in this scenario involves loading each 3-D object instance into the scene. Repeated loading of the same object in different locations can lead to inefficiencies.

   To address this, future developments in component twins could involve a system where a single instance of a 3-D object is streamed, loaded into memory, and rendered multiple times as needed. This approach would optimize the handling of 3-D objects in digital twin environments, enhancing efficiency and reducing the computational load.

2. 2)

   Asset twins: This methodology focuses on creating data-driven digital twins using a library of physics-based reduced-order models. When a single model library is shared among numerous assets, this approach can effectively scale to applications requiring a substantial number of digital twins (Krzyczkowski 2019). Asset twins involve an estimation process wherein online sensor data from a physical asset determines which models from the library should be integrated into the digital twin. Future advancements in asset twins should enhance the robustness of model selection, particularly in the context of corrupted data. Implementing mechanisms to improve robustness and incorporating various damage models to detect and classify actual asset damage is also essential. GE Healthcare (Kapteyn et al. 2020) has noted the application of asset twins in healthcare, addressing challenges such as staffing model design and surgical block schedule optimization.

3. 3)

   System twin: Operating at a higher level, system twins amalgamate different assets to form a complete functional system, such as a vehicle's brake system (Aghdam et al. 2021). These twins offer insights into asset interactions, thereby augmenting overall performance.

4. 4)

   Process twin: Process twins utilize high-performance computing to optimize equipment and manufacturing processes. This is achieved by integrating multidimensional process knowledge models (Aghdam et al. 2021). Manufacturers can attain unparalleled efficiency and deeper insights by combining production processes with economic considerations.

5. 5)

   Application: A digital twin system integrating Virtual Reality (VR) and Artificial Intelligence (AI) technologies has been developed to monitor and analyze welder behavior. This system exemplifies the practical application of digital twin technology in understanding and improving specific work processes.

The transformation of physical assets into digital twins involves an in-depth asset data collection process, which is then utilized to form an exact digital counterpart. This procedure is essential for asset management and predictive maintenance. There are variant data models and datasets used to underpin the digital twin initiative and significantly enhance the effectiveness and capabilities of digital twin implementations while reducing development efforts and optimizing the total cost of ownership. Many software applications have recently been used to create and manage digital twins. This section presents samples of Data models, Datasets, and software applications.

5.1 Smart city data models and datasets

To illustrate the potential of digital twins in smart cities, let us consider examples of digital twin data models and datasets that provide valuable insights for urban planning and management. Digital twin data can be applied in both tangible and virtual realms. These data are pivotal for asset monitoring, operational optimization, and safety enhancement in physical settings. On the other hand, virtual landscapes enable realistic simulations, training endeavors, and strategic planning. This dual use of digital twins highlights their adaptability, effectively bridging the real and digital domains.

One of the cornerstones of DT design and development is modeling data. Data originate from heterogeneous sources, use various protocols, and include their own data attributes, attribute types, and relationships. In order to ensure interoperability, it is necessary not only to standardize the communication between DT components but also to standardize the data format that flows through these components.

3D city modeling transcends mere data acquisition and processing, extending into data management, storage, and exchange. Consequently, open and standardized data models and exchange formats are essential for 3D city modeling. CityGML and its streamlined counterpart, CityJSON (Ledoux et al. 2019), are the most established data formats for 3D city models. These formats facilitate representations ranging from basic to richly detailed, depending on the required level of detail (LoD). The building model is depicted in five levels of detail, from LOD0 to LOD4, with higher LoDs offering more detail and accuracy. The aim is to manage the complexity of 3D models effectively.

In their study, the authors (Lei et al. 2022) assess 40 authoritative 3D city models that have emerged since 2013. This evaluation yields both quantitative and qualitative insights. The framework developed offers a thorough and structured comprehension of the landscape of semantic 3D geospatial data while also serving as an evaluated compilation of open 3D city models.

In (Ledoux et al. 2019), digital twin (DT) initiatives in cities are classified based on the nature of their digital replicas (static or dynamic, i.e., incorporating sensor or IoT data) and the extent of data integration (the data connection between the physical and digital worlds). Various static datasets utilize digital model integration, including Helsinki 3D + , Espoo, Vienna, Zurich 3-4D, and Amsterdam3D. Meanwhile, dynamic datasets such as Digital Twin Munich, Rennes 3D, Virtual Gothenburg, and Sofia-Bulgaria employ digital shadow integration. Furthermore, dynamic datasets like DUET, Fishermans, and Virtual Singapore implement digital twin integration. It can be inferred that most initiatives are digital shadows, given that data connections from the real world to the digital copy are automated. At the same time, the reverse typically involves manual processes (human interventions adapting the physical world). This bidirectional connection warrants further exploration.

Many research projects and similar initiatives mainly focus on collecting and providing IoT data generated from smart cities. For example, the ODAA platform (2016)^{Footnote 1} provides open access to data collected from the City of Aarhus using IoT infrastructure deployed within the city. The datasets within the ODAA are categorized across various applications, including energy, population and society, transport, education, and more. Moreover, San Francisco Open Data (2024)^{Footnote 2} and the City of Chicago Data Portal (2024)^{Footnote 3} provide a centralized collection of relevant smart city datasets that are publicly accessible.

For example, the NYC Open Data Initiative has already leveraged digital twin technology to improve urban planning and citizen engagement. By providing access to a wide range of open data, including information on infrastructure, public services, and environmental factors, the initiative has empowered citizens to actively participate in shaping the city's future.

5.2 Software for digital twin creation and management

Numerous digital twin software applications are available for creating and managing digital twins in buildings, cities, and urban systems. Some notable examples include:

• Autodesk revit (Autodesk 2019): This software is extensively used for Building Information Modeling (BIM) and is acclaimed for its comprehensive design, documentation, and collaboration tools. It enables architects, engineers, and construction professionals to create detailed 3D models and provides extensive data for informed decision-making throughout a building's lifecycle.

• Esri cityengine (Badwi et al. 2022): CityEngine is a robust software tool for crafting 3D city models. It is utilized by urban planners and designers to generate detailed and lifelike representations of cities, offering capabilities for cityscape generation, urban environment modeling, and simulation of various urban scenarios. It also integrates with GIS data to enhance city models with geographic information and analysis.

• Bentley systems openbuildings designer (Mainisa et al. 2023): This BIM software provides advanced building design and construction modeling tools. Architects, engineers, and construction professionals use it for detailed 3D modeling, structural analysis, and effective collaboration throughout the building lifecycle.

• Unity reflect (Nämerforslund 2022): Unity Reflect is a platform that creates interactive and immersive experiences with digital twins. It supports real-time, high-fidelity 3D modeling for virtual and augmented reality environments, enhancing visualization, interaction, and decision-making processes.

• Siemens city performance tool (Al-Obaidy et al. 2022): Specifically tailored for urban planning and management, this tool offers a comprehensive platform for analyzing and optimizing the performance of urban systems.

• iLens from knowledge lens: This leading Industrial IoT solution addresses Industry 4.0 needs with capabilities in Interface Connectivity, Edge Computing, Monitoring and Control, and Predictive Analytics. iLens is powering diverse industries globally, including Automation, Manufacturing, Energy, and Utilities.

• Iotics: Iotics' innovative digital twin technology enables seamless communication across an entire digital ecosystem. It bridges gaps between various entities, from sensors to power stations and individual trains to entire airplane networks, transcending organizational boundaries and differing data languages while maintaining security.

• Kavida.ai: This supply chain digital twin platform assists enterprises in making intelligent resiliency decisions. It builds supply chain digital twins using artificial intelligence to help enterprises prevent and mitigate disruptions in real time or before they occur.

• MODS reality: This cloud-based application hosts a digital twin of a facility in a point cloud environment, enhancing engineering and streamlining scheduling and work execution management for maintenance and minor modifications, thereby maximizing performance and profitability.

• Twinzo: As a mobile-first live digital twin platform focused on operational excellence, twinzo visualizes and reconstructs live data in 3D, offering novel ways to analyze and consume information. It helps customers save significant operational costs and increase production output.

• VEERUM's digital twin: This application is a leading visualization and analytics tool that combines CAD, geospatial, document management, IoT, and operational systems. It delivers considerable cost and time savings in operations, maintenance, reliability, and complex capital construction projects.

• WillowTwinTM: Revolutionizing the built world, WillowTwinTM is a pioneering software platform for real estate and infrastructure assets. It provides a central hub for all asset data, turning siloed datasets into a virtual replica of the built form. The platform enables proactive, data-driven decision-making in real-time to reduce costs, increase profits, and manage risks.

Extensive research has been conducted on digital twins (DTs) and their applications, yet a standard method for assessing DT performance remains elusive. Establishing a method for evaluating the performance of DTs is essential for enhancing or monitoring processes and systems within a business context. Such a method could guide researchers and practitioners in developing more effective digital twins (Psarommatis and May 2022).

There have been limited studies focusing on specific methodologies for assessing DT performance. Chen et al. (Chen et al. 2021) proposed a DT maturity model for managing industrial assets based on Gemini principles, facilitating quantitative evaluation of DT flexibility and implementation levels. Chakraborty et al. (Chakraborty and Adhikari 2021) assessed DT performance in a multi-time scale dynamical system using an efficient framework that leverages expectation maximization and a sequential Monte Carlo sampler for developing machine learning-based DTs. Shangguan et al. (Shangguan et al. 2020) evaluated DT performance for fault diagnosis using a predefined threshold technique, focusing on accuracy (ACC), specificity (SPE), and sensitivity. Psarommatis et al. (Psarommatis and May 2022) introduced a systematic approach for measuring DT performance and flexibility, quantifying it based on four key performance indicators (KPIs). Additionally, they introduced DTflex as a new KPI to evaluate the flexibility of digital twins.

6.1 Performance metrics categories

Although there are no well-established methods or Key Performance Indicators (KPIs) in the field for thoroughly assessing the performance of Digital Twins (DT), this study suggests classifying performance metrics according to three essential elements: software, hardware, and data management middleware. This paradigm makes it possible to evaluate the system's efficacy in detail. A thorough analysis of the body of prior research and industry norms guided the choice of these indicators. We aimed to find measures that captured the essential elements of DT performance by combining knowledge from several sources.

The proposed metrics ensure an adequate evaluation by focusing on DT performance characteristics within each component. For example, metrics about hardware components evaluate attributes like scalability, communication dependability, and sensor precision. These metrics were selected to represent the fundamental hardware performance features essential to DT's operation. Similarly, metrics related to middleware for data management emphasize security, scalability, and efficiency, highlighting middleware's vital role in integrating and controlling data streams. Finally, software component metrics highlight the significance of strong software functions for DT performance by addressing factors such as model integrity, simulation accuracy, and user interface responsiveness. Each metric recommended in this section is supported by its relevance to real-world DT implementations and alignment with broader business or operational objectives. These measures help stakeholders make well-informed decisions by offering practical insights about DT performance. Including these measures also attempts to create a standard framework for assessing DT performance in various applications and domains.

6.1.1 Hardware components

• Sensor accuracy: Precision and reliability of physical sensors.

• Communication reliability: Efficiency of data transmission between sensors and the digital counterpart.

• Hardware scalability: Ability to expand hardware components with increasing data volumes.

• Latency in data acquisition: Time taken to acquire and transmit sensor data.

• Hardware failure rate: Frequency and severity of failures in sensors or actuators.

6.1.2 Data management middleware

• Data integration efficiency: Ease of integrating data from various sources into the DT.

• Middleware latency: Time taken for middleware processes to complete tasks.

• Data accuracy and consistency: Precision and consistency in data storage and management by middleware.

• Scalability of middleware: Ability to handle increasing data volumes without performance degradation.

• Data security protocols: Effectiveness of security protocols in protecting data during storage and transit.

6.1.3 Software components

• Model fidelity: Accuracy and completeness of the digital model representing the entity.

• Simulation accuracy: Precision of simulations compared to real-world scenarios.

• Quality of visualization: Clarity and detail of visual representations in the user interface.

• User interface responsiveness: Speed and responsiveness of the software interface to user actions.

• IoT device integration: Compatibility and integration with various IoT devices.

• Scalability of software: Capacity to handle increasing computational loads and data processing demands.

• Software security: Protections against cyber threats and unauthorized access.

• Interactivity and control: Responsiveness of software to user inputs and control commands.

• Updating and maintenance efficiency: Ease of updating and maintaining software components.

• Effectiveness of decision support: Capability of the software to provide meaningful insights.

6.2 Best practices for evaluating digital twin performance

As noted by Peter Drucker, Mgt. consultant and author, “You cannot manage what you cannot measure.” This principle is equally applicable to digital twins. The confusion matrix employed in data science can measure digital twins' performance ARC Advisory Group (2024).^{Footnote 4} Assessing the performance of digital twins necessitates a thorough approach that considers multiple aspects, including hardware, data management middleware, and software components. Below are some essential practices for effectively assessing the performance of digital twins. The formulation of these best practices necessitated a thorough examination of the current literature on DT performance evaluation. Consultations with some stack holders were also conducted. By combining information from various sources, we hoped to convey the multidimensional nature of DT performance and provide meaningful advice to practitioners and researchers alike. Furthermore, the methods were iteratively refined to ensure their usefulness and applicability across various situations and industries.

Each practice recommended in this section is based on known management principles and its ability to address important difficulties in DT performance evaluation. For example, the emphasis on objective definition and particular Key Performance Indicators (KPIs) demonstrates the significance of goal alignment and measurement precision in achieving effective DT efforts. Similarly, data quality, security assessment, and scalability analysis methods emphasize these variables' importance in assuring the dependability and efficacy of distributed computing systems.

• Objective definition: Clearly articulate the goals and objectives of the digital twin implementation to align performance indicators with broader business or operational objectives.

• Establish specific KPIs: Identify and set specific Key Performance Indicators (KPIs) that align with the objectives, ensuring they are measurable, relevant, and linked to desired outcomes.

• Multidimensional evaluation: Assess performance across multiple dimensions, including accuracy, responsiveness, scalability, security, and usability.

• Regular review and update of metrics: Given the evolving nature of digital twin environments, performance metrics should be regularly reviewed and updated to maintain relevance and accuracy.

• Focus on data quality and integrity: Emphasize metrics related to data accuracy, consistency, and integrity, as the quality of the digital twin largely depends on the reliability of its data.

• Incorporate end-user experience metrics: Include metrics that gauge user satisfaction and adoption, such as visualization quality, interaction, and ease of use.

• Measure latency and responsiveness: Evaluate latency in data collection, middleware processing, and software responsiveness to ensure real-time or near-real-time capabilities.

• Security performance assessment: Implement metrics to evaluate the efficacy of security measures, including data encryption protocols.

• Scalability analysis: Examine the digital twin's scalability, focusing on how well it accommodates increasing data volumes, user numbers, and processing requirements.

• Simulation accuracy verification: Regularly validate the accuracy of simulations and virtual representations against actual world scenarios to ensure the digital twin's reliability.

• Benchmarking: Compare performance against industry standards or best practices to understand how the digital twin stacks up against similar implementations.

• Utilize monitoring technologies: Deploy monitoring technologies that offer real-time insights into the digital twin's operation, enabling proactive issue identification and resolution.

• Develop a continuous improvement process: Establish a process for continuous improvement that integrates user feedback and ongoing evaluations, fostering a culture of perpetual enhancement.

By adhering to these practices, organizations can establish a robust framework for assessing and improving the performance of their digital twins, ensuring that these technologies deliver maximum value and effectively contribute to strategic objectives. To sum up, this section's performance indicators are the outcome of a systematic approach guided by academic research and industry observations. We hope to give readers a thorough grasp of how these measures support efficient DT evaluation procedures by outlining the reasoning behind their selection and their applicability to DT performance assessment.

Understanding the obstacles encountered while deploying digital twin technology is critical for its successful adoption and improvement. This section elucidates the difficulties various components of digital twin systems face, shedding light on their origins and implications. The challenges outlined are meticulously identified through an extensive review of literature and insights from field and industry experts, signifying their significance in the successful deployment and operation of digital twin systems. This analysis integrates multiple sources to pinpoint these hurdles as key challenges. The study organizes the identified challenges into three main aspects of digital twin technology: hardware, data management middleware, and software. This categorization facilitates a thorough understanding of the complex problems impacting different aspects of digital twin systems. A thorough examination of these challenges across the hardware, data management middleware, and software components aids in bridging the current research gap. Whereas prior studies often discussed these challenges in broad strokes, (Tuhaise et al. 2023) divided them into three categories: data transmission, interoperability, and data integration. This research details specific problems within each distinct component of the digital twin framework, thereby offering an in-depth analysis of the inherent obstacles in digital twins. It identifies hardware-related challenges, such as the complexity of sensor integration and issues with hardware reliability, suggesting solutions like adopting standardized sensor interfaces and employing predictive maintenance strategies. Furthermore, the study uncovers problems in data management middleware, including data integration bottlenecks and interoperability issues, recommending developing scalable middleware systems and adopting universal standards to enhance interoperability. The research outlines security vulnerabilities and algorithmic complexity regarding software components, proposing using advanced analytical tools and robust cybersecurity measures as solutions.

By delineating these issues across hardware, middleware, and software components, the study enhances the understanding of digital twin technology and offers actionable recommendations for enhancing the technology’s effectiveness and resilience. As digital twin technology continues to evolve, the findings underscore the necessity of concentrating on these components to surmount challenges and fully exploit the technology's potential across various applications and industries. The examination of digital twin elements and their associated challenges is visually summarized in Fig. 13, which consists of three parts: (a) delineates the components of a digital twin, (b) identifies the challenges specific to each component, and (c) proposes solutions to these challenges.

Overview of digital twin components, associated challenges, and proposed solutions. Part (a) delineates the core components of a Digital Twin (DT). In part (b), a detailed breakdown highlights the challenges inherent in each component. Part (c) provides insightful solutions strategically proposed to address these challenges and enhance the effectiveness of Digital Twin implementation

Full size image

7.1 Hardware components

Hardware components are the foundation of digital twin systems, comprising sensors, actuators, and other physical devices. Challenges within this component include:

• Sensor integration complexity: Integrating diverse sensors for real-time data poses compatibility and synchronization issues.

• Hardware reliability: Ensuring long-term sensor and actuator reliability is essential.

Proposed solutions involve adopting standardized sensor interfaces and implementing predictive maintenance strategies to mitigate these challenges.

7.2 Data management middleware

Middleware plays a crucial role in managing and processing the vast amount of data generated by digital twin systems. Challenges within this component include:

• Data integration bottlenecks: Handling diverse data streams can lead to processing delays.

• Interoperability issues: Different standards may hinder middleware system compatibility.

Proposed solutions include developing scalable middleware architectures and embracing industry-wide standards for improved interoperability.

7.3 Software components

Software components encompass the algorithms and analytical tools for real-time data analysis and decision-making. Challenges within this component include:

• Algorithmic complexity: Complex algorithms for real-time analytics and decision-making need streamlining.

• Security vulnerabilities: Software components are susceptible to cybersecurity threats.

Proposed solutions involve utilizing advanced analytical tools and robust cybersecurity protocols to address these challenges.

In conclusion, addressing the challenges linked with digital twins requires a deep understanding of their core components: hardware, data management middleware, and software. This analysis has unveiled various obstacles, from hardware constraints to data integration complexities and software interoperability challenges. A comprehensive perspective is provided by examining these issues across the distinct hardware, middleware, and software components. It is essential to identify and tackle the limitations associated with hardware, the challenges within middleware, and the issues related to software interoperability to enhance the efficiency and robustness of digital twin systems. As digital twin technology evolves, prioritizing these areas will be critical for navigating difficulties and leveraging the technology’s capacity in diverse applications and industries.

Case studies in the realm of smart cities and digital twins serve as vital illustrations of these technologies in practical scenarios:

• Dubai's "Happiness Agenda": A smart city initiative using big data to enhance urban living and measure "happiness" across various criteria. The objective was to involve every citizen in shaping future cities, particularly focusing on citizen engagement. Dubai’s "Happiness Agenda" implementation represents a notable example of a smart city involving its residents in urban development. Dubai has positioned itself as one of the "happiest" places to live by defining citizen "happiness" across multiple criteria. It uses big data analysis to allocate urban resources strategically, enhancing the city's overall "Happiness Index" (Zakzak 2019).

• West Cambridge site and IFM building: These case studies explore adaptable digital twins at the building level, integrating various data sources and AI-driven decision-making. The West Cambridge site of the University of Cambridge in the UK was chosen as a case study due to its diverse facilities, which include university buildings, sports centers, residence areas, main roads, parking places, and restaurants. This variety allows for testing and evaluating the proposed dynamic digital twin system across different types of infrastructure. Additionally, the site's size and complexity offer an ideal environment to assess the effectiveness of the technology. Access to extensive data sources, collaboration opportunities with experts, and relevance to the academic community further contribute to its suitability as a testbed for the study (Qiuchen Lu et al. 2019).

• Herrenberg, Germany: A case study demonstrating the use of digital twin technology in urban planning and city management. The case study of Herrenberg might illustrate the implementation and benefits of digital twins in improving urban planning, infrastructure management, and citizen engagement within the city. Herrenberg was selected as a case study for the digital twin due to its relevance to urban challenges, accessibility of diverse data sources, the potential for collaboration with local stakeholders, engagement of the community, and suitability in terms of size and complexity for testing the digital twin technology (Dembski et al. 2020).

• Cambridge Sub-region: A digital twin pilot is developed, integrating diverse data streams for urban planning and decision-making. The authors stress the significance of including diverse data like IoT sensors, satellite images, social media, and government records to ensure an all-encompassing and precise city representation. The case study presented in the paper is about developing a digital twin pilot for the Cambridge Sub-region. It highlights how integrating various data streams and simulation models can assist urban planning, resource allocation, and decision-making processes. The case study provides insights into the potential benefits of using a city-level digital twin for improving efficiency, sustainability, and resilience in urban environments (Wan et al. 2019).

• Málaga City: Implementing cognitive analytics in smart city management to enhance transportation, energy, and public services. The focus is on enhancing various aspects of urban life, such as transportation, energy management, waste management, and public services. The case study of Málaga City demonstrates the practical implementation of cognitive analytics to improve decision-making processes, optimize resource allocation, and ultimately enhance the quality of life for its residents (Pérez and Toledo 2017).

• Ålesund, Norway: The study explores the role of a data-driven digital twin in enhancing urban systems and services within a smart city framework. It suggests using high-quality 3D graphical digital twins (GDTs) of cities to generate 4D visualizations of geolocalized time-series data to enhance citizen engagement. Through a case study conducted in Ålesund, Norway, the methodology utilizes readily available hardware and a game engine to develop immersive environments for presenting complex data sourced from GIS, BIM, demographics, and IoT. The approach emphasizes scalability, transferability, versatility in data integration, adherence to privacy regulations, and dependable data delivery. The paper introduces a pioneering smart city GDT framework, which capitalizes on interactive features and advancements in metrology (Major et al. 2021).

• Case study in Greece: Details the development and application of digital twins tailored for smart cities, focusing on urban infrastructure improvements. The case study probably illustrates how digital twins optimize city systems, improve efficiency, and facilitate decision-making processes in Greek urban environments. This study might showcase practical examples of implementing digital twin technology to address challenges and enhance the overall functioning of a smart city in Greece (Evangelou et al. 2022).

Each case study offers unique insights into the deployment and impact of digital twin technology in various urban settings, highlighting its potential to improve city management and living standards. Table 2 offers an overview of each paper's focus areas, case studies, and key highlights, showcasing their distinct contributions and applications in the field of digital twins in smart cities.

In the evolving discourse on smart cities and digital twin technologies, a critical examination of multi-level governance, organizational practices, and governance dimensions emerges as pivotal. The collective contributions from the referenced studies provide a comprehensive overview of the challenges and strategies in implementing smart city initiatives across different governance frameworks and geographical contexts.

As examined in one study, the integration of Chinese new authoritarian principles into smart government transitions highlights the inherent tensions between state-level directives and local-level implementation, underscoring the complexity of multi-level governance in authoritarian regimes (Zhang and Mora 2023). This perspective is enriched by a nuanced exploration of organizational practices within smart city development, revealing how bureaucratic, technocratic, and participatory logics intersect to shape decision-making and citizen engagement in smart city projects (Mora et al. 2023a). Furthermore, the identification of three key governance dimensions—institutional context for urban innovation, urban innovation ecosystem, and urban digital innovation—provides a framework for understanding the governance mechanisms essential for fostering smart city transitions (Mora et al. 2023b).

Critical analysis across the studies reveals common challenges in smart city governance, such as interoperability and compatibility issues within the digital ecosystem and integrating a technological dimension in urban development. These challenges underscore the importance of addressing interoperability and compatibility to enhance city planning and management effectively (Quek et al. 2023). The discourse extends to the critical analysis of smart urbanism in non-Western contexts, notably in India and Africa, where issues of urban informality, equity, and the inclusivity of smart city initiatives are brought to the forefront (Prasad et al. 2023; Tonnarelli and Mora 2023). These analyses highlight the necessity of adopting equitable and inclusive smart city development approaches that consider the needs and priorities of all urban dwellers, particularly marginalized communities.

Moreover, the call for empirical studies and the integration of innovation management theory into smart city governance research emphasizes the need for practical guidance and theoretical advancements in managing urban digital innovation (Mora et al. 2023b). The exploration of human-cyber-physical interactions further illuminates the evolving relationship between technology, governance, and societal dynamics, advocating for a holistic approach that balances technological advancements with ethical and sociocultural considerations (Quek et al. 2023).

In conclusion, the amalgamated insights from these studies advocate for a pragmatic, contextually informed, and inclusive approach to smart city governance. By addressing the multifaceted challenges of interoperability, governance, and citizen engagement, and by incorporating a critical perspective on urban informality and inclusivity, this body of work contributes significantly to the scholarly discourse on smart cities and digital twins. The emphasis on empirical research, innovation management, and the integration of technology in urban development underscores the dynamic interplay between technology, governance, and urban development strategies in the quest for sustainable and equitable urban futures.

9.1 Role of DT in smart city governance

Smart city governance constitutes a complex framework fundamental to the effective realization and long-term viability of smart city endeavors. It encompasses the strategic alignment of policies, technological systems, and multifaceted collaborations amongst stakeholders by overarching urban development goals. Digital twin technology plays a pivotal role in enhancing smart city governance by offering innovative solutions across various components:

• Policy and strategy formulation: Crafting policies and strategies that guide smart city initiatives in service of the city's broader objectives (Beckers 2022). Digital twins assist in crafting policies and strategies by providing valuable insights derived from real-time data and simulations. City authorities can utilize digital twins to assess the impact of different policies and strategies on urban systems, enabling informed decision-making aligned with broader city objectives.

• Collaborative ecosystem: Fostering partnerships spanning government entities, the private sector, academic institutions, and the citizenry, thus leveraging collective knowledge and resources (Beckers 2022). Digital twins foster collaboration among government agencies, private sector entities, academic institutions, and citizens by providing a platform for data sharing and analysis. This collaborative ecosystem enhances collective knowledge and resource utilization, facilitating more effective governance practices and co-creating solutions to urban challenges.

• Technological infrastructure: Establishing and administering the technological basis, encompassing data management and digital platforms, that underpins smart city operations (Zhang and Mora 2023). As a foundational element of smart city operations, digital twins contribute to establishing and managing the technological infrastructure required for governance. They enable comprehensive data management and visualization, empowering city administrators to monitor urban systems, identify emerging trends, and respond proactively to issues in real-time.

• Ethical considerations: Prioritizing ethical concerns by safeguarding data privacy and security and ensuring the equitable deployment of technology (Mora et al. 2023a). Digital twins support ethical governance by prioritizing data privacy, security, and equitable technology deployment. Through robust data encryption protocols and access controls, digital twins safeguard sensitive information, ensuring that governance processes remain transparent, accountable, and inclusive for all stakeholders.

• Public participation: Stimulating citizen involvement in the governance process promotes transparency and inclusiveness (Mora et al. 2023b). Digital twins facilitate public participation by providing accessible platforms for citizen feedback, collaboration, and co-design of urban solutions. By incorporating citizen inputs into decision-making processes, digital twins help ensure that governance strategies align with community needs and preferences.

• Sustainability: Championing sustainable development practices integrated within smart city projects to prioritize environmental stewardship and long-term resilience (Quek et al. 2023). By simulating various scenarios and assessing the environmental impact of proposed policies and projects, digital twins enable city authorities to prioritize sustainability and resilience in urban planning and decision-making processes.

9.2 Smart city governance challenges

The pursuit of smart city objectives is frequently hindered by governance crises, underscoring the complexities of managing urban digital transformations. Digital twins offer innovative solutions to navigate the complexities of urban governance and enhance decision-making processes. Here, we explore how digital twins can be utilized to tackle key challenges in smart city governance:

• Data privacy and security concerns: Contending with data privacy and security risks associated with the vast collection and storage of urban data (Mora et al. 2023a). Digital twins incorporate robust data encryption protocols and access controls, ensuring the protection of sensitive information within smart city systems. Digital twins help mitigate privacy and security risks associated with urban data collection and storage by enabling secure data management and transmission.

• Digital divide and inequity: Mitigating the digital divide can potentially intensify social disparities within urban communities (Prasad et al. 2023). Digital twins promote inclusivity and bridge the digital divide by providing accessible platforms for citizen engagement and participation in governance processes. Through user-friendly interfaces and interactive visualization tools, digital twins empower all citizens to contribute to decision-making, regardless of their technological literacy or socioeconomic status.

• Regulatory and legal challenges: Navigating the disparity between the swift pace of technological progress and prevailing regulatory frameworks (Tonnarelli and Mora 2023). Digital twins assist city authorities in navigating regulatory and legal frameworks by providing comprehensive data analytics and scenario modeling capabilities. Digital twins facilitate informed policy-making and ensure alignment with legal standards and industry regulations by simulating the impact of proposed regulations and assessing compliance requirements.

• Fragmented governance structures: Surmounting the intricacies of multi-stakeholder governance structures, which can obstruct coordinated action (Zhang and Mora 2023). Digital twins serve as centralized platforms for data integration and collaboration, overcoming the challenges of fragmented governance structures. By consolidating diverse datasets from multiple stakeholders and domains, digital twins enable seamless information sharing and coordination, fostering synergy among various governmental entities and stakeholders.

• Resource constraints: Confronting limitations in financial, technical, and operational capacities is vital to the success of smart city ventures (Quek et al. 2023). Digital twins optimize resource utilization and operational efficiencies within smart city governance through predictive analytics and optimization algorithms. By identifying inefficiencies and optimizing resource allocation, digital twins help cities overcome resource constraints and maximize the impact of limited financial, technical, and operational resources.

By integrating digital twin technologies, smart city administration can address these challenges, paving the way for innovative solutions and sustainable urban development. Digital twins offer a comprehensive and data-driven approach to governance, enabling cities to enhance decision-making processes, accountability, and transparency, ultimately enhancing the quality of life for urban residents. While digital twin technologies have the potential to significantly improve urban management through advanced data analytics, simulation, and optimization, their seamless integration into smart city governance requires careful consideration of governance issues. This includes addressing concerns related to data privacy, fostering collaboration among stakeholders, and upholding ethical principles.

This survey paper employs a meticulous bibliometric methodology, selecting the Web of Science database for its comprehensive coverage and developing precise search criteria to gather over 4,220 relevant articles. The analysis uses advanced tools like VOSviewer for network analyses and visualizations, including co-authorship and keyword co-occurrence maps, enabling a detailed examination of trends and relationships in Digital Twin technology and Smart Cities research. This methodological rigor ensures the study's reliability and contributes to its uniqueness in the field. This survey comprehensively reviews over 4200 publications in the domain of Digital Twins and Smart Cities. It outlines the evolution, applications, and integration of Digital Twins with IoT and AI in urban development. The survey distinguishes itself through extensive bibliometric analysis, focusing on datasets, platforms, software, and performance metrics, and it offers unique insights into the challenges and opportunities within the field. The findings include emerging trends, key thematic areas, and a detailed exploration of various Smart City applications. The paper concludes with implications for urban developers, policymakers, and researchers and recommendations for future research directions. The field of Digital Twin (DT) and Smart Cities is ripe for future research, aiming to overcome current challenges and explore new frontiers. Detailed investigation and development in this area are essential for realizing the full potential of DT technologies in urban environments. The discussions pave the way for sustainable and equitable urban futures, recognizing the dynamic interplay between technology, governance, and urban development strategies.

Future research should focus on:

• Enhanced data integration: Developing more efficient methods for integrating diverse data sources within DT systems.

• Scalability solutions: Creating scalable DT models suitable for larger and more complex urban environments.

• Advanced security protocols: Strengthening cybersecurity measures for DT systems to ensure data privacy and security.

• Sophisticated analytical tools: Incorporating cutting-edge AI and machine learning techniques for predictive analytics and decision-making.

• Expanding IoT capabilities: Extending the use of IoT in DTs for comprehensive real-time data collection and monitoring.

• Sustainable urban development: Leveraging DTs for resource management, focusing on sustainability and environmental conservation.

• Citizen engagement models: Developing DTs prioritizing citizen involvement in urban planning and management.

• Policy and governance studies: Examining the influence of policy in guiding DT implementation and addressing ethical concerns.

• Economic impact assessment: Evaluating the economic implications of DTs, including cost analysis and return on investment.

• Real-world case studies: Documenting extensive case studies to assess DTs' practical impact and challenges in urban settings.

• Investigating future technological advancements: new applications, and the role of policy and governance in Digital Twins development.

• Summarizing the main findings and implications and a call to action for further research in this evolving field.

The findings of this paper are poised to influence future research, policy-making, and practical applications in Smart Cities and Digital Twins in significant ways:

• Informing future research directions: The comprehensive review of over 4,200 publications provides valuable insights into the current state of Digital Twins and Smart Cities research. Researchers can utilize this information to identify gaps in existing literature and prioritize areas for further investigation. For example, identifying challenges such as data integration bottlenecks and security vulnerabilities can guide future research efforts toward developing solutions to these pressing issues.

• Guiding policy development: Policymakers can leverage the findings of this paper to inform the development of policies and regulations related to Digital Twin technology and its application in Smart Cities. By understanding the challenges and opportunities associated with Digital Twins, policymakers can create frameworks that promote innovation while addressing data privacy, cybersecurity, and ethical considerations.

• Improving urban planning and management: The insights provided by this paper can assist urban planners and city managers in making informed decisions about adopting and implementing Digital Twins in Smart Cities. By understanding Digital Twin technology's potential benefits and challenges, city officials can develop strategies to optimize urban infrastructure, improve resource management, and enhance citizen services.

• Driving technological innovation: The paper identifies emerging trends and technological advancements in Digital Twin technology, such as the integration of AI and IoT, as well as the development of scalable models and advanced security protocols. These insights can inspire innovation in academia and industry, leading to the development of new tools, platforms, and solutions that push the boundaries of Digital Twin technology and its applications in Smart Cities.

Finally, the findings of this article have the potential to spark advances in research, policymaking, and practical applications connected to Digital Twins and Smart Cities, resulting in more efficient, sustainable, and resilient urban development.