Research progress and prospect of digital twin in bridge engineering

Abstract

The concept of digital twins in bridge engineering is still vague and even confused with the Bridge Information Model (BrIM). Therefore, this study provides a detailed review of 42 papers related to digital twins in bridge engineering, focusing on a proper definition, key features and creation techniques for bridge digital twin (BDT). The paper also compares BDT and BrIM from the perspectives of their elements, features, fidelity, services provided, and degree of development. The applications of BDT at different life cycle stages are identified, and the related technologies are analyzed in detail. The results show that the research clusters of BDT are divided into geometric model generation, finite element model updating, and management and are focused on the operation and maintenance phase while lacking attention in the design and construction phase. Besides, a reference framework of BDT based on the life cycle of bridges is proposed, and directions for future research are suggested.

Keywords

bridge digital twin bridge information model bridge design optimization intelligent bridge construction life cycle bridge maintenance

Introduction

Bridges serve as crucial components of civil infrastructure, responsible for both transportation and public safety. However, many bridges have surpassed their design life, leading to an increase in damages and failures as they age (Zhao et al., 2022). This poses significant challenges for their maintenance and management. Bridge projects have been relatively slow in adopting digitalization, particularly intelligent technologies (Opoku et al., 2021). Fortunately, the emergence of advanced technologies such as building information model (BIM), big data, artificial intelligence (AI), and digital twin (DT) has made it possible to realize digital, green, and smart bridges, which represent the future direction of bridge engineering. BIM has gained widespread use in the construction industry due to its visualization capabilities and coordination advantages. In bridge engineering, the application of BIM, known as Bridge Information Model (BrIM), is progressively gaining traction (Costin et al., 2021; Lee et al., 2012). The digital twin is a rapidly evolving technology that involves the use of digital models to replicate physical entities (Grieves, 2014). It is considered to be a core technology for Industry 4.0, driving the development of smart manufacturing. For example, it can be employed to enhance product design, manufacturing processes, and service delivery (Tao et al., 2018). The production plan can be simulated, evaluated and improved in the virtual world, and the real-time data collected from the physical world can be compared with the plan to achieve rational and accurate manufacturing management (Qi and Tao, 2018).

The use of digital twins in civil engineering is still at the early stage compared to manufacturing and aerospace industries (Pregnolato et al., 2022). The building industry has shown greater progress in integrating BIM, internet of things (IoT), and sensor data to develop dynamic digital twin demonstrators for buildings (Vivi et al., 2019). In contrast, the adoption of both BIM and digital twin technologies in bridge engineering lags behind, with the concepts of BrIM and digital twin not yet clearly distinguished. For example, the digital twin is considered the same as a BrIM that aims to digitize and inform various physical and functional features of a project (Kaewunruen et al., 2021). Therefore, it is necessary to understand how the digital twin is defined in bridge engineering. Specifically, how it is implemented and how it changes the design, construction, operation, and maintenance of bridges throughout their life cycle.

In this study, we conducted a comprehensive review on the applications of digital twins in bridge engineering, focusing on five key research questions:

How to define the digital twin in bridge engineering and what are its features?

What are the differences between digital twin and BrIM?

What technologies are involved in digital twin and how to implement a digital twin in bridge engineering?

What applications can be realized using digital twin in bridge engineering?

What are the major challenges of the current digital twin technologies in bridge engineering and what are the future research?

This study aims to clarify the concept and applications of digital twins in bridge engineering, providing valuable insights for researchers and practitioners. The methodology employed in this study is presented. Then, it provides a discussion on the definition, characteristics, and relationship between the digital twin and BrIM in bridge engineering. Following that, the applications of digital twins in bridge engineering and associated technologies are explored, and an implementation framework is proposed. Besides, the challenges and future directions of digital twins in bridge engineering are discussed. Finally, the conclusions of this study are summarized.

Research methodology

A systematic literature review, which is widely used to conduct literature reviews (Hallaji et al., 2022; Rathore et al., 2021), was employed in this study to conduct an extensive search and detailed analysis to answer the research questions presented in the previous section. The process of selecting publications using this method was implemented based on the preferred reporting items for systematic reviews (PRISMA) guideline that is a set of criteria established for the quality of research in systematic reviews and provides minimum standards that must be fulfilled for review articles (Page et al., 2021). The procedure of using the systematic review is illustrated in Figure 1.

Figure 1.

Implementation process of systematic literature review.

Keyword searching

The Scopus and Web of Science databases were selected for literature searches. The former is faster in indexing and updating papers than the other databases, while the latter is more advantageous in searching the literature over a longer period (Falagas et al., 2008). A broader set of search strings was used for subject search (including title, abstract, and keywords) to maintain the precision and breadth of the literature search. Meanwhile, the type and time frame of the literature was set to be unlimited to avoid missing publications and to maintain research diversity. The results showed that 287 documents were retrieved from the Web of Science database and 339 from the Scopus database by June 2023.

Articles screening

As the documents retrieved in the previous step would not exactly match the topic of this study, they were further manually screened by reading their titles and abstracts. The criteria for manual screening of publications are as follows:

The articles are presented in English and have clear authors.

The research covers both bridges and digital twins, including concepts, methods, applications, etc.

Patents and magazines are selected based on their research value by reading the full text.

Conference papers are only included if the conference has been held for more than 10 consecutive sessions and the content of the conference paper by the same author is not similar to that of the journal paper.

Documents retrieved from the Web of Science with the above screening criteria were reduced from 287 to 32 valuable papers, and all of them were journal papers. Similarly, 339 publications in the Scopus database were filtered by the above criteria and 42 papers were selected, including 36 journal papers and 6 conference papers. Finally, after combining all the screened papers and eliminating duplicates, a total of 42 papers (see Table 1) were included in this study, including 36 journal papers and 6 conference papers. Compared with a review on digital twin applications in the construction industry with 22 papers (Opoku et al., 2021), the number of 42 papers obtained in this study is still advantageous.

Table 1.

Statistics of Papers Related to Digital Twins in bridge Engineering.

Ref.	Cita	Journal/Conference	Types
(Hu et al., 2023)	-	Advances in civil engineering	E
(Lu and Brilakis, 2019)	101	Automation in construction	E
(Broo et al., 2022)	18	Automation in construction	F, E
(Pregnolato et al., 2022)	9	Automation in construction	F
(Song et al., 2023)	3	Automation in construction	F
(Yang et al., 2022)	2	Automation in construction	E
(Pantoja-Rosero et al., 2023)	1	Automation in construction	E
(Gao et al., 2023)	-	Automation in construction	F, O&M
(Bittencourt et al., 2021)	1	Proceedings of the 10th international conference on bridge maintenance, safety and management, IABMAS 2020	F, O&M
(Lai et al., 2022)	4	Case studies in construction materials	O&M
(Lu et al., 2020)	1	Construction research congress 2020	E
(Febrianto et al., 2022)	3	Data-centric engineering	E
(Sánchez-Rodríguez et al., 2020)	6	EG-ICE 2020 workshop on intelligent computing in engineering, proceedings	E
(Jiang et al., 2021b)	28	Engineering structures	O&M
(Guo and Fang, 2023)	-	Engineering with computers	E
(Broo and Schooling, 2021)	11	IEEE access	F
(Dang et al., 2022)	22	IEEE transactions on industrial informatics	F, O&M
(Dan et al., 2021)	12	IEEE transactions on intelligent transportation systems	F, O&M
(Shim et al., 2019b)	15	International conference on smart infrastructure and construction 2019	O&M
(Van Nimmen et al., 2021)	7	Journal of bridge engineering	O&M
(Omer et al., 2021)	7	Journal of bridge engineering	E, O&M
(Zhao et al., 2022)	7	Journal of bridge engineering	O&M
(Ye et al., 2020b)	26	Journal of civil structural health monitoring	O&M
(Baisthakur and Chakraborty, 2021)	7	Journal of civil structural health monitoring	O&M
(Dang and Shim, 2020)	14	Lecture notes in civil engineering	O&M
(Bello et al., 2022)	3	Lecture notes in civil engineering	F, O&M
(Zhou et al., 2022)	-	Lecture notes in civil engineering	F, O&M
(Kang et al., 2021)	27	Multimedia tools and applications	F, O&M
(Mohammadi et al., 2021)	37	Remote sensing	E
(Marra et al., 2021)	3	SCIRES-IT	E
(Shao et al., 2020)	19	Sensors	E
(Ghahari et al., 2022)	6	Sensors	O&M
(Yu et al., 2022)	5	Structural control and health monitoring	O&M
(Yoon et al., 2022)	2	Structural and multidisciplinary optimization	O&M
(Ye et al., 2019)	40	Proceedings of the 12th international workshop on structural health monitoring	O&M
(Shu et al., 2019)	3	Proceedings of the 12th international workshop on structural health monitoring	E, O&M
(Jiang et al., 2021a)	4	Structural monitoring and maintenance	O&M
(Shim et al., 2019a)	99	Structure and infrastructure engineering	F, O&M
(Omer et al., 2019)	45	Structure and infrastructure engineering	E, O&M
(Ye et al., 2020a)	7	Structure and infrastructure engineering	O&M
(Jiang et al., 2022a)	1	Structure and infrastructure engineering	F, O&M
(Futai et al., 2022)	3	Structure and infrastructure engineering	F, O&M

Note: The number of citations in the table is based on the Scopus data as of June 2023. The meaning of the symbols in the table is defined in the following section.

Bibliographic analysis

The research topics, journal distribution, and core keywords of all selected papers were analyzed based on the bibliometric analysis method. First of all, as shown in Figure 2, digital twin research in bridge engineering emerged in 2019 and experienced significant growth in 2021, indicating a growing interest in this field. On the other hand, it is not limited to civil engineering journals but extends to various fields, including computer science and sensors. This interdisciplinary nature highlights the complexity of integrating digital twin technology into bridge engineering.

Figure 2.

The number of annual publications of digital twins in bridge engineering.

The 42 papers selected are further divided into two categories of digital twin creation and applications based on the research theme of the papers and the life cycle of the bridge. The former includes the development framework (denoted by F) and model establishment (E) of the digital twin, while the latter includes the design (D), construction (C), and operation and maintenance (O&M) of the bridge’s life cycle, as detailed in Table 1. It can be observed that the majority of digital twin applications in bridge engineering concentrate on the operation and maintenance phase. This emphasis can be attributed to the dynamic connection facilitated by bridge health monitoring, which enables effective utilization of digital twin. Additionally, in terms of citation count, a paper on the automatic generation of geometric digital twins for reinforced concrete bridges from point clouds garnered the highest number of citations by June 2023 (Lu and Brilakis, 2019). A close second was a paper exploring the application of digital twins in the O&M stage of small and medium span bridges (Shim et al., 2019a). These findings indicate that current digital twin research in bridge engineering primarily focuses on the geometric modeling of digital twins and their applications in bridge O&M.

The cluster analysis of co-occurrence keywords was performed based on the references in Table 1. The visual bibliometric network was mapped to identify the knowledge structure and components of the digital twin in bridge engineering. As shown in Figure 3, the size of the nodes is proportional to the frequency of keyword appearance, while different colors represent distinct clusters. The research clusters of digital twins in bridge engineering were found to fall into three categories: existing bridge geometric model generation, bridge finite element model updating, and bridge management and maintenance. These clusters demonstrate that the application of digital twins in bridge projects is still primarily focused on traditional aspects such as bridge health monitoring and numerical simulations. However, it also reveals that digital twins are being utilized throughout the entire life cycle of bridges, rather than being limited to a specific phase. The top ten buzzwords identified through keyword co-occurrence analysis for digital twins in bridge engineering include life cycle, structural health monitoring (SHM), maintenance, decision making, deterioration, inspection, finite element method, information management, BrIM, and model updating. These buzzwords affirm that the current emphasis of digital twin applications in bridge projects lies in inspection, monitoring, and decision-making during the operation and maintenance phase. It suggests that digital twins are proving to be a compelling tool in enhancing these aspects of bridge management.

Figure 3.

Keyword co-occurrence analysis visualization network.

The relationship between the digital twin and BrIM in bridge engineering

In this section, the definitions and characteristics of digital twin and BrIM will be summarized separately, and their similarities and differences will be compared. Also, this section serves as a detailed answer to Questions 1 and 2 raised in the previous section.

Definition and development of digital twin

The concept of the digital twin has emerged in the past two decades, and a standardized definition has yet to be established. Tao et al. (2019a) outlined the development of the digital twin in three stages: formation, incubation, and development, based on the number of related papers published over the years (refer to Figure 4).

Figure 4.

Development trend of the digital twin research (Tao et al., 2019a).

The initial appearance of the digital twin concept can be traced back to Grieves' presentation on product life cycle management (PLM) (Grieves and Vickers, 2017). Grieves envisioned PLM to encompass real space, virtual space, and data or information connectivity between the two. However, the concept of the digital twin remained relatively ambiguous during this stage. In 2011, The National Aeronautics and Space Administration (NASA) introduced the digital twin concept for predicting the life of vehicle structures and presented its formal definition, marking the first practical application of digital twin (Glaessgen and Stargel, 2012). Concurrently, the manufacturing industry, closely tied to aerospace, began exploring the use of digital twin. For example, a digital twin of a real machine can be built on a cloud platform and use the collected data and existing knowledge to simulate health conditions (Lee et al., 2013). Despite these advancements, research on digital twin was still relatively limited, and application development was hindered by technological constraints related to communication, information, and sensors.

The first white paper on the digital twin was published in 2014 (Grieves, 2014), signifying the expanding scope of digital twins into diverse industries such as energy, healthcare, and transportation. During the subsequent rapid development stage, different industries provided their versions of the digital twin definition, reflecting variations in characteristics and goals (Liu et al., 2021; Negri et al., 2017; Opoku et al., 2021). Grieves and Vickers (2017) defined the digital twin as a set of virtual information structures that represent actual or potential physical products, enabling the inspection of physical products through the digital twin. Furthermore, Rathore et al. (2021) defined the digital twin as a process involving the creation of a cyber twin in the digital world that corresponds to a physical entity, process, or system. Despite the varied definitions, certain fundamental elements are presented in the digital twin concept. These include a physical entity in the real world, a virtual twin replicating the physical entity in a virtual space, bidirectional data or information flow between the physical entity and the virtual twin, and the provision of services by the virtual twin equivalent to those offered by the physical entity.

Definition, characteristics, and modeling of bridge digital twin

The concept of bridge digital twin has emerged to represent the implementation of digital twins in bridge engineering. However, a unified definition of bridge digital twin is still lacking, and their characteristics and roles remain unclear. A conceptual model divided the digital twin of a steel bridge into three parts: the physical steel bridge system, the virtual steel bridge system, and the data or information that connects the two, based on the original concept of the digital twin (Jiang et al., 2021a). In another perspective, the bridge digital twin is considered a technique to create virtual models of bridges, simulate their function and performance, and analyze data for real-time evaluation and predictive maintenance (Yu et al., 2022). Additionally, some studies identify BrIM or finite element models as digital twins. For example, Meixedo et al. (2022a, 2022b) equated the bridge finite element model with the digital twin when validating a damage detection method for railroad bridges. Kaewunruen et al. (2021) defined bridge digital twin as BrIM, aiming to digitize and inform various physical and functional characteristics of the project.

Based on these established definitions, as illustrated in Figure 5, a bridge digital twin (BDT) can be understood as a digital, informational, and intelligent process for bridge design, construction, and management. It involves collecting real bridge data and environmental information to create virtual twins of physical bridges in the digital world and establishing interactions between them to enable real-time or near real-time updating and evolution of both throughout their lifecycle. Ultimately, virtual bridges are used to provide services that can be considered as services provided by physical bridges, such as damage detection, real-time monitoring, dynamic analysis, life prediction, and intelligent management. It is noted that the bi-directional interaction between physical and virtual bridges is a distinctive characteristic of BDT. The acquired sensor data from the physical twin flows to the virtual twin to keep the two synchronized, and the data or information from the virtual bridge, such as historical information, simulation results, and predicted parameters, can in turn stream to the physical bridge for adjusting instrument deployment or monitoring component changes. Meanwhile, the virtual twin is bi-directionally connected to the service system, receiving data like mission commands and expert knowledge from the service system as well as transmitting the resulting analysis, simulation and other data to the service system. The data aggregated by the service system is fed back to the physical bridge to achieve specified tasks, such as management decisions, capacity, and maintenance activities of the physical bridge. In summary, the main features of BDT include:

BDT is a multi-element integrated system encompassing physical and virtual bridges, bi-directional data flow, and associated services.

BDT can be implemented throughout the whole bridge’s life cycle.

The virtual bridge serves as a real-time or near real-time representation of the physical bridge, capturing its geometric information, semantic information, structural performance, and health status.

The data of the physical bridge updates the virtual bridge synchronously, whilst the virtual bridge can influence the physical bridge by adjusting construction schedule, sensor locations, etc.

Virtual bridges can provide intelligent services that represent the state and performance of the proposed or existing bridges, such as visual design, virtual construction, real-time monitoring, what-if scenario analysis, and preventive maintenance.

Figure 5.

Components of bridge digital twin.

Currently, bridge digital twins are modeled in a variety of ways and also at discrete and smaller time scales. For example, digital twin models of bridges in the design phase are represented directly in computer-aided design (CAD) models (Guo and Fang, 2023). The geometric digital twin (gDT), which reflects the variation of real bridge geometric features, and the numerical digital twin (nDT), which simulates the real structural behaviour and performs simulation analysis, are the more mentioned ones. The predominant method for creating gDT is to acquire bridge point cloud data by laser scanning and manually classify them based on geometric features (Sánchez-Rodríguez et al., 2020). Efforts have been made to improve the accuracy and efficiency of generating the gDT from point cloud data. A slice-based object fitting method was proposed to automatically generate gDT of reinforced concrete bridges from labeled point clusters (Lu and Brilakis, 2019). However, this method currently faces challenges in accurately and automatically generating bridges with complex geometries. Similarly, a system was developed for the rapid generation of gDT by automatically identifying target areas and corresponding scan paths, although it still requires manual processing of large amounts of data (Lu et al., 2020). Another study employed a deep learning-based weighted super point graph (WSPG) method for automatic semantic segmentation of bridge components from full-size bridge point clouds (Yang et al., 2022). However, the method still struggles with consistent boundary distinction between adjacent components. There are also works on how the geometric information of the bridge can be accurately extracted from the arch bridge with missing point cloud to generate the gDT (Hu et al., 2023). Photogrammetry has also been employed to create the gDT of bridges (Pantoja-Rosero et al., 2023; Zhou et al., 2022). A comparison between unmanned aerial vehicle (UAV) photogrammetry and terrestrial laser scanning (TLS) for generating bridge point clouds revealed that TLS-based models offer higher geometric accuracy and point density, while the UAV-based method is advantageous in terms of implementation time and equipment cost (Mohammadi et al., 2021).

On the other hand, finite element model updating, as a digital twin technique (Ye et al., 2020a), is widely used to create the nDT of bridges for real-time analysis, prediction, and simulation. Although numerical models for structural analysis can be directly generated from the gDT of bridges, the accuracy and real-time capability of nDT created using this method require improvement (Shu et al., 2019). An alternative approach to creating the nDT involves updating structural parameters using Bayesian inference methods and monitoring response data (Febrianto et al., 2022; Ghahari et al., 2022). Additionally, a combination of computer vision and vibration measurement has been utilized to achieve geometric and stiffness property updates of the nDT of a bridge (Lai et al., 2022). However, this method has yet to be validated on in-service bridge structures.

Definition, characteristics, and modeling of BrIM

BrIM is a term used in bridge engineering to describe the application of BIM specifically to bridges (Lee et al., 2012). While BIM is defined by the International Organization for Standardization (ISO) as the use of a shared digital representation of building assets for decision making, it is important to note that BrIM and BIM are not equivalent due to the differences in structural components, data patterns, and modeling approaches between bridges and buildings. According to the Federal Highway Administration (FHWA), BrIM is a comprehensive digital representation of the physical and functional characteristics of a bridge, utilizing standardized digital formats to enable various digital processes and decision making throughout the bridge's life cycle (Chipman et al., 2016). This is also the definition of BrIM adopted by this study.

BrIM has become a powerful tool throughout the bridge’s life cycle (Huang et al., 2011). In the design phase, it has revolutionized traditional two-dimensional (2D) CAD drawings by enabling three-dimensional (3D) parametric and visual design. This improves efficiency and quality, as demonstrated by Girardet and Boton who utilized an improved parametric modeling approach for bridge design and analysis (Girardet and Boton, 2021). BrIM also facilitates collaboration among stakeholders and reduces time and costs in subsequent bridge construction. During the construction phase, BrIM enhances collaboration and communication, reducing conflicts (Biancardo et al., 2020) and enabling the simulation of construction scenarios in a 3D virtual space to determine optimal plans (Lee et al., 2012; Li et al., 2012). BrIM can also be used for cost estimation to control project capital investment (Marzouk and Hisham, 2014). In the operation and maintenance phase, BrIM's rich 3D information, data sharing, and information exchange capabilities improve the efficiency and proactivity of bridge management (Rashidi and Karan, 2018). McGuire et al. (2016) used BrIM to connect and analyze data related to bridge inspection, evaluation, and management, providing maintenance recommendations. Sacks et al. (2018) proposed the SeeBridge system which integrates bridge inspection defects with BrIM for rapid inspection and assessment of bridges. Isailović et al. (2020) developed a bridge management system based on BrIM to accurately reflect the current condition of bridges.

Despite the advantages of BrIM in geometric visualization, data integration, and management informatization, there are still technical barriers and functional limitations in its use throughout the bridge life cycle. These include difficulties in transferring and sharing information due to various data formats and schemas used in BrIM. The reliability, accuracy, and functionality of BrIM depend on the level of development (LoD) of the bridge model, and achieving the highest LoD500 with rich semantic information and sensing data can be challenging (Cheng et al., 2016). BrIM models are typically developed with different LoDs for specific applications and are not continuously updated throughout the bridge's life cycle. Finally, BrIM cannot convey information or data to the real bridge or facilitate automatic evolution, limiting its capability for bi-directional flow of information (Khajavi et al., 2019).

The creation of BrIM has transitioned from research to practical implementation, supported by various commercial software such as Autodesk Revit, Tekla Structures, and Bentley OpenBridge Modeler. Designers have successfully utilized these tools to directly create BrIMs. For example, Biancardo et al. (2020) employed the Revit software to design the "IV Bridge" at Naples Capodichino International Airport. Girardet and Boton (2021) utilized parametric algorithms on the Tekla Structures platform to construct BrIMs and perform structural analysis. Additionally, it can be generated based on 2D bridge design drawings. Akanbi and Zhang (2022) developed a semi-automatic method to convert 2D bridge plans (PDFs) into 3D information models using the Revit platform and generating industry foundation classes (IFC) files. However, their approach lacks full automation in generating BrIMs and does not associate semantic information with bridge components. Moreover, techniques such as laser scanning (Yang et al., 2020) and photogrammetry (Rashidi and Karan, 2018) can be employed to create 3D information models of existing bridges, similar to the creation of geometric digital twin (gDT).

Comparison of BDT and BrIM

According to the above analysis, the similarities and differences between BDT and BrIM are observed as follows:

Physical bridges are essential for BDT development, but not for BrIM.

BDT requires a higher fidelity level of the virtual bridge, aiming for full comprehensibility (Jones et al., 2020), while BrIM can be photorealistic (Cheng et al., 2016).

BDT enables bi-directional data flow between the physical and virtual bridges, whereas BrIM only allows data transfer from the physical to the virtual bridge.

The virtual bridge of BDT has an inheritance in each phase of the whole life cycle, which differs significantly from the virtual bridge in BrIM.

Both BDT and BrIM can create digital models of existing bridges using laser scanning and photogrammetry. Additionally, BrIM can be used to build initial virtual bridges to enable 3D visualization design and virtual construction during the design and construction phases when physical bridges do not yet exist.

Both BDT and BrIM provide basic services like 3D visualization, virtual assembly, construction simulation, cost estimation, and maintenance decisions. However, BDT offers additional intelligent services such as real-time monitoring, dynamic analysis, scenario simulation, and management decisions throughout the bridge's life cycle.

Both BDT and BrIM integrate various advanced technologies, including IoT, sensing, cloud computing, laser scanning, augmented reality, virtual reality, big data, and AI.

BDT is still in the exploratory stage of theory and applications, while BrIM has reached commercial implementation.

The overall comparison between BDT and BrIM is provided in Table 2. In summary, BDT and BrIM are different approaches to address challenges in traditional bridge design, construction, and O&M. While BrIM focuses on simulating the physical and functional characteristics of a bridge at a specific point in time, BDT provides a real-time or near real-time virtual twin that continuously represents changes in the physical bridge throughout its life cycle. BDT can also integrate BrIM technology to digitally transfer data and information from the bridge design to the construction phase. Therefore, BDT offers a more advantageous and imaginative solution for bridge design, construction, and O&M.

Table 2.

The Similarities and Differences Between BDT and BrIM.

Categories	BDT	BrIM
Virtual bridge	Yes, and real-time twinning with a physical bridge	Yes, not linked
Physical bridge	Required	Not required
Features	Data flows bi-directionally, self-evolves, and spans the entire life cycle	Unidirectional data flow, static model
Highest fidelity	Replicas	Photorealistic effects
Services	Intelligent services throughout the bridge’s life cycle	Specific services such as 3D visualization and maintenance
Maturity	Budding stage	Commercially available

Application and framework of bridge digital twin

This section provides a detailed analysis and summary of the applications of BDT throughout the bridge’s life cycle. As shown in Figure 6, the applications using BDT in the entire life cycle can consist of two applications in the design phase, three applications in the construction phase, and six applications in the O&M phase. In addition, the architecture of the BDT is discussed to provide a generic and feasible reference framework.

Figure 6.

Applications of BDT in the entire life cycle of bridges.

Design phase

Bridge design encompasses various stages, including conceptual design, preliminary design, and detailed design, each requiring continuous refinement. Traditional design methods, relying on 2D drawings and experiential knowledge, present challenges in communication among project participants and increase the risk of rework. However, digital twins offer the potential to integrate information from stakeholders, the surrounding environment, and component materials, thereby aiding bridge design (Jiang et al., 2021c). Moreover, digital twinning provides a traceable digital process throughout the product's lifecycle, enabling designers to have a comprehensive digital footprint for continuous optimization and informed decision-making during different design phases (Tao et al., 2019b).

During the design phase, the physical bridge does not yet exist, making it difficult to perform data acquisition and mapping operations to create a bridge's digital twin based on its physical entity. Nonetheless, the integration of digital twin technology in bridge design can be facilitated by leveraging existing advanced technologies. For example, the BrIM creates a 3D information model, enabling visualization of internal bridge data. Additionally, the geographic information systems (GIS) complements BrIM by providing environmental and geographic data related to the bridge project, such as lighting, terrain, and temperature (Göçer et al., 2016). The ongoing development of technologies like big data, cloud computing, and AI further expands the potential value of digital twins in bridge design. It should be noted that studies on digital twins at the bridge design stage have not been found in this literature (see Table 1). This study attempts to develop an exploratory analysis of the application of the digital twin to the bridge design process from a visual design and verification optimization perspective.

Visualization design

The 3D visualization provided by the digital twin enables a "what you think is what you get" pattern for bridge design and provides a fully reliable decision-making tool. Visualization design helps designers to improve the quality and efficiency of bridge design. It also enables project stakeholders to check whether the designed bridge meets the requirements in a perceptible and experiential virtual environment and provides direct feedback to the designer, which enhances collaboration and reduces unnecessary rework on bridge projects.

The implementation of visualization design for BDT involves a variety of advanced technologies. First of all, BrIM and GIS technology can be integrated to display bridge 3D information, external environment, and geographic information in real-time during the bridge design process, which will provide strong support for designers in determining the spatial location, structure, and construction plan of the bridge. Besides, some emerging technologies such as virtual reality (VR), augmented reality (AR), and mixed reality (MR) are used to generate corresponding bridge virtual environments for 3D visualization of the review experience. Although no papers have been published on digital twins related to bridge design in bridge engineering, the value of digital twins for visualization design has been proven in other fields. For example, in a highway widening project, a digital twin of the underpass road and a BIM model of the newly widened road can be created during the preliminary design phase to help with clearance checks and redesign (Jiang et al., 2022b). Therefore, the development of advanced technologies and algorithms should be promoted to realize the visualization design of BDT.

Verification optimization

During the design phase, virtual bridges that represent the performance of physical bridges can be effectively verified and optimized without consuming much time, money and labor. Therefore, the verification optimization of BDT is crucial to improve the quality and efficiency of bridge design. Virtual verification provides the opportunity to identify and eliminate delivered bridge models that are not ideally aligned with design goals and inconsistent with specification requirements (Grieves and Vickers, 2017). Specifically, virtual verification involves both expectation checking and performance testing. To begin with, the difficulty of bridge design lies in balancing the demands of project stakeholders. BrIM technology can be exploited to check and verify whether the designed bridge model meets the pre-agreed requirements of project stakeholders, which can avoid design rework during the construction phase. On the other hand, the virtual twin designed by BrIM and GIS technologies is simulated to check the bridge performance under various application scenarios. The success of bridge design comes from continuous improvement. The virtual bridge of BDT can predict the performance of the physical bridge to indicate the direction of design optimization, which can efficiently achieve iterative optimization of bridge design. In addition, a BDT with data traceability features can provide a complete and continuous design footprint to help designers record and read every stage in the design and verification process.

Construction phase

The construction of bridges faces various challenges, including large components, complex work environments, and lengthy construction periods. Unlike the highly automated and informationized manufacturing of industrial products, bridge construction has not reached the same level of advancement. However, many new technologies and methods have been researched and used to solve the problems of high construction costs, coarse site management, high safety risks, and poor construction quality in the bridge construction process. The implementation of BrIM technology in bridge construction within the Denver metropolitan area was found to reduce change orders and rework, resulting in a 5%–9% reduction in construction costs (Fanning et al., 2015). A risk visualization and information management method based on Risk Breakdown Structure (RBS) and 3D/4D BIM technology was proposed for site risk identification and control in bridge projects (Zou et al., 2019). A material delivery monitoring system, utilizing closed-circuit television (CCTV) and radio frequency identification (RFID) technologies, enabled systematic management of construction materials (Ju et al., 2012). These efforts have helped to improve the informationization and intelligentization of bridge construction to a certain degree. However, while these technologies and methods address specific problems in the construction process, they cannot serve as the primary tool for seamless data updates throughout the entire construction process. For example, BrIM technology cannot interact dynamically with real bridges due to its data being static.

As a solution, a BDT emerges as a promising approach for achieving intelligent bridge construction. By simulating the entire construction process of a bridge in a digital environment before its physical realization, BDT enables evaluation of project constructability and optimization of construction plans. Furthermore, technologies like sensing, communication and laser scanning can be leveraged to acquire data and information about the ongoing construction and the surrounding environment, facilitating updates to the virtual bridge within the BDT. Additionally, the real-time connection of personnel, materials, machinery and virtual bridges is realized through RFID and IoT to promote intelligent management of construction sites.

Virtual construction

With the emergence of complex terrain conditions, innovative construction methods, and oversized member assemblies in bridge construction, simulating the entire construction process in a digital environment has become crucial. Lee et al. (2012) used BrIM to assemble components in a 3D virtual space to find interferences between components and errors in design and to optimize the assembling sequence and path of components in advance. Virtual prototyping simulation (VPS) has also been adopted to simulate different construction scenarios of bridges to avoid conflicts between construction equipment and construction elements. However, the limitation of VPS is that it cannot determine the optimal construction plan (Li et al., 2012).

In contrast, the BDT is the best tool to realize virtual construction as it can simulate bridge construction considering dimensions, construction schedule, cost, quality, safety and other multi-dimensional information, which helps construction planners to optimize the allocation of resources, cost, time, and quality. The virtual construction of BDT can predict the performance of bridge structures during construction and identify potential problems that are overlooked by traditional simulations to help review and optimize the constructability of bridge projects. The visualization of the entire assembling process can help train construction workers to improve assembling efficiency and quality.

Construction monitoring

Several case studies have emerged in the field of on-site monitoring for bridge construction. A wireless real-time monitoring system was developed to enhance the efficiency of bridge constructions by identifying and measuring the human posture of construction workers during operations (Bai et al., 2012). Integration of an nD computer-aided design (CAD) object and a telepresence concept video management operation system allowed monitoring of the current construction conditions of the bridge (Kang et al., 2016). The vision and IoT sensors were employed to monitor the behavior of construction workers working at height, aiming to prevent fall accidents (Khan et al., 2022). However, a BDT with its advanced technology integration and continuous data interaction capability offers a promising approach for real-time monitoring of all elements at the bridge construction site. Firstly, technologies such as sensing, laser scanning, computer vision, and communication can acquire real-time information about the physical bridge and the surrounding environment at the construction site. Secondly, by comparing the physical bridge under construction with the virtual bridge design, differences in data and information can be identified, facilitating updates to the virtual bridge within the BDT. Lastly, a continuously updated virtual bridge enables efficient assessment of construction quality, accurate identification of hazards at the construction site, and real-time monitoring of construction progress.

On-site management

The management of resources, including construction workers, component materials, and mechanical equipment, is a complex and crucial task in bridge construction. While the development of a prefabricated building field service platform has enabled intelligent management of construction schedule and cost using IoT, RFID, BIM, and VR technologies, it falls short in achieving comprehensive management of safety, quality, and the environment (Li et al., 2018). Therefore, there is a great potential and need to leverage BDT to improve the intelligent management of bridge construction sites. It can effectively break the barriers between information on materials, labor, equipment, quality, and safety to support interaction and collaboration among all-element, as well as integrate technologies such as BrIM, RFID, IoT, and CCTV to support various management activities (Opoku et al., 2021). Furthermore, it facilitates a digital workflow, ensuring complete recording and traceability of all data and information generated during bridge construction, including material supply, equipment condition, and personnel operations. This enables project managers to make efficient and accurate decisions. However, the implementation of digital twins for intelligent management is still in the conceptual stage, and limited case studies exist due to the complexities of the construction site and information technology constraints.

Operation and maintenance phase

The operation and maintenance (O&M) phases of bridges constitute a significant portion of their entire life cycle and require substantial resource investment. Bridge owners and operators are particularly concerned about the safety, intervention needs, and remaining service life of bridges (Ye et al., 2020a). However, the data and information related to the bridge construction are usually not completely transferred to the bridge managers due to the fact that the bridge builders are not involved in the subsequent operation and maintenance. Moreover, traditional inspection techniques struggle to identify hidden defects that are inaccessible to personnel or equipment (Omer et al., 2019; Ye et al., 2020a). Also, it is difficult to make a comprehensive and accurate performance prediction of bridges through traditional SHM. To this end, various methods and technologies have been proposed to address the challenges at the O&M stage, focusing on bridge perception, evaluation, and decision-making. The BrIM was used to develop bridge health monitoring systems that can organize and visualize a significant amount of sensor data and subsequent structural health information (Li et al., 2022). It can also be integrated with MR technologies for remote bridge inspection and maintenance work (Nguyen et al., 2021). UAV photogrammetry and machine learning algorithms were employed to automatically identify damage patterns on bridges (Morgenthal et al., 2019). However, these individual technologies and methods have not yet achieved the establishment of an integrated management and maintenance platform for bridge perception, evaluation, and decision-making.

Therefore, as shown in Table 3, researchers have increasingly recognized the value of applying BDT at the O&M stage, aiming to establish an integrated management and maintenance platform. First of all, real-time data and information on in-service bridges are acquired using measurement, scanning, and monitoring devices, enabling dynamic connection between virtual bridges and their physical counterparts. This allows virtual inspection and real-time monitoring of the actual bridges. The nDT of the bridge is continuously updated for remote analysis and diagnosis using model updating methods and bridge monitoring data. Additionally, the nDT of the bridge, combined with intelligent algorithms, facilitates performance prediction of the bridge structure. Furthermore, BDT can perform scenario simulation to complete “what-if” analysis that cannot be performed on physical bridges, which will help provide decision-making support to bridge managers. Finally, advanced technologies can be integrated to enable intelligent actions such as security alerts, automatic control, proactive maintenance, and retrofit reminders.

Table 3.

The applications of BDT at the O&M stage.

Categories	Applications	References
Perception	Virtual inspection	(Marra et al., 2021; Omer et al., 2019, 2021)
Perception	Real-time monitoring	(Dang et al., 2022; Zhou et al., 2022; Ye et al., 2019; Shao et al., 2020; Lai et al., 2022)
Evaluation	Analysis diagnosis	(Ye et al., 2020b; Ghahari et al., 2022; Yu et al., 2022; Dan et al., 2021; Shu et al., 2019; Yoon et al., 2022; Lai et al., 2022)
	Performance prediction	(Yu et al., 2022; Jiang et al., 2021a, 2021b; Zhao et al., 2022)
	Scenario simulation	(Baisthakur and Chakraborty, 2021; Van Nimmen et al., 2021; Jiang et al., 2022a)
Decision-making	Intelligent actions	(Dan et al., 2021; Bittencourt et al., 2021; Dang and Shim, 2020; Bello et al., 2022; Jiang et al., 2022a; Futai et al., 2022; Shim et al., 2019a)

Virtual inspection

The virtual inspection of BDT relies on data from laser scanners, cameras, and sensors to update the gDT of bridges. This enables non-contact, visual, and dynamic inspections in a virtual environment. For example, Marra et al. (2021) created a gDT of a historic stone arch bridge using laser scanning and photogrammetry, facilitating virtual inspections. Similarly, Omer et al. (2019, 2021) employed light detection and ranging (LiDAR) scanning to generate a gDT and developed a virtual reality application for bridge inspection. These cases demonstrate that BDT is becoming an important and promising method for bridge inspection.

However, there are still some pressing issues that need to be investigated in BDT-based virtual inspection. First of all, the lack of comprehensive digital defect information from the beginning of bridge service leads to insufficient type and accuracy of virtual inspection to identify defects. A sustainable and updatable digital twin of the bridge should be established in the future from the perspective of the whole bridge’s life cycle. On the other hand, the twinning rate between virtual and physical bridges in the gDT of bridges is still low and needs to be improved to enhance the virtual inspection in real time.

Real-time monitoring

BDT provides a virtual representation of real bridges, continuously reflecting their condition through sensors and machine learning. The condition data of the virtual bridge can be used to calibrate the monitoring data of the physical bridge to help determine the type of damage. To date, the use of BDT for condition monitoring is still in its infancy, focusing on establishing health monitoring frameworks. Dang et al. (2022) developed a digital twin framework (cDTSHM) for structural health monitoring, achieving 92% accuracy in damage detection. Also, Zhou et al. (2022) proposed a digital twin framework based on drones, cameras, and accelerometers for long-term and non-disruptive bridge health monitoring. Ye et al. (2019) developed a data-driven and model-based digital twin system for bridge health monitoring, offering efficient data processing and interpretation.

The twinning rate between physical and virtual bridges in BDT is a crucial research direction for real-time and accurate bridge health monitoring. Computer vision techniques are being explored to capture bridge information (Lai et al., 2022; Shao et al., 2020), but validation on real bridges is needed.

Analysis diagnosis

The condition analysis and diagnosis based on BDT is the best realization to assess the safety condition of bridges. Ye et al. (2020b) demonstrated the importance of developing 3D digital twins to identify root causes of bridge defects, exemplifying this with a 30-year-old highway bridge. A BDT can provide real-time and historical data for comprehensive analysis, compensating for data deficiencies and handling unknown anomalies. Updating the nDT of bridges enables the assessment of structural conditions, and various approaches have been explored. Ghahari et al. (2022) proposed continuous Bayesian model updating to modify the bridge's finite element model (FEM) for evaluating damage during seismic events. Yu et al. (2022) analyzed stress response, considering vehicle loads and temperature effects on the fatigue damage state, by updating the digital twin numerical model with data from the dynamic weighing and SHM systems. Dan et al. (2021) employed measured bridge traffic load data for online mechanical response analysis.

Moreover, there is growing interest in creating nDT directly from as-is models for structural analysis. Shu et al. (2019) conducted linear finite element analysis using 3D point cloud data, focusing on bridge structural performance. However, The point cloud focuses more on the appearance of the bridge structure and does not consider the degradation of material properties over time. Yoon et al. (2022) used a UAV to check the material property degradation of bridge structural members to update the FEM for seismic fragility analysis. Lai et al. (2022) combined computer vision 3D reconstruction and vibration measurements to assess bridge structural condition using heterogeneous information sources. Although the feasibility of this method for in-service bridges is uncertain, it allows for updating geometric and stiffness properties of the FEM.

In summary, BDT-based analysis and diagnosis of bridge safety conditions primarily involve FEM updating, relying on monitoring data to correct structural modal properties. These advancements enhance bridge safety assessment and pave the way for further research in the field.

Performance prediction

Bridge performance prediction is crucial for quantifying future service performance, preventing unexpected failures, and aiding decision-making. Researchers have focused on utilizing BDT, which combines multi-source data and numerical models to improve prediction reliability and accuracy. Yu et al. (2022) predicted fatigue life in cable-stayed bridges by updating BDT with data from dynamic weighing and SHM systems. Jiang et al. (2021a, 2021b) forecasted fatigue life in steel bridges using BDT, allowing real-time interaction between physical and virtual models. However, this method is limited to segmental welded joints with U-shaped ribs.

BDT can also be used to predict the structural performance of bridges. Zhao et al. (2022) used long-term monitoring data to update transverse distribution factors and influence lines, enabling real-time performance prediction. However, the accuracy and applicability may be reduced by assuming axle weights can be assigned to main girders based on the transverse distribution factor. Although BDT-based performance prediction shows promise, challenges remain due to data availability and the level of abstraction required for updates. Further research is needed to fully exploit the potential of BDTs in bridge engineering.

Scenario simulation

BDT provides a virtual twin that is equivalent to a real bridge in real-time, which allows performance evaluation to be verified on bridges in a virtual environment as an alternative to field load tests that may cause irreversible damage to the bridge. Baisthakur and Chakraborty (2021) developed digital twins for validation load tests using an improved Hamiltonian Monte Carlo model updating method, reducing costs and risks. Van Nimmen et al. (2021) created a digital twin of a pedestrian bridge to simulate its dynamic performance under crowd-induced loading accurately. BDT can be used to simulate the performance evolution of a bridge under predefined maintenance measures to help bridge managers make better decisions and draw up maintenance plans. Jiang et al. (2022a) simulated the actual fatigue crack expansion process throughout the life cycle of the steel bridge based on BDT to obtain the optimal detection, monitoring, and repair strategy. While FEM updating is currently the main approach for BDT-based scenario simulation, the focus should be on building comprehensive virtual equivalents that represent the bridge's conditions and exploring BDT-based simulations from a holistic life cycle perspective.

Intelligent actions

The intelligent actions of BDT are the process of making unified and integrated decisions on the operation and maintenance of bridges through the massive amount of perception data (including real-time data and historical data) and the evaluation results of the virtual twin that is equivalent to the physical bridge in real-time. These intelligent actions include security warning (Dan et al., 2021), proactive maintenance (Bittencourt et al., 2021; Dang and Shim, 2020), and asset management (Jiang et al., 2022a; Bello et al., 2022). Furthermore, the BDT represents the real behavior of the bridge as well as its future performance, which can help engineers specify a long-term strategy for bridge operations management (Shim et al., 2019a). For example, Futai et al. (2022) built a digital twin of a 6-span reinforced concrete railroad bridge to better support bridge operators in decision-making and unified management. Shim et al. (2019a) proposed a digital twin-based maintenance system for small and medium span bridges, combining a continuously updated digital twin model with a digital inspection system. While these BDT-based cases show promise, their effectiveness in real bridge projects requires further testing. Additionally, the development of a comprehensive and standardized technical framework for BDT-based intelligent actions is necessary.

Research on digital twin framework for bridges

Several frameworks have been proposed for developing digital twins in bridge engineering. A fatigue whole life cycle management framework for steel bridges was presented to support optimal strategies for detection, monitoring, and repair (Jiang et al., 2022a). The framework consists of a physical bridge system, a virtual bridge system, and their linkage. Bridge operation information is collected through the physical system, and fatigue-related data is transferred to the virtual system for real-time calibration and updates. Pregnolato et al. (2022) extended the framework by adding an experience system to the physical, virtual, and connected systems. Dang et al. (2022) incorporated a cross-platform user application deployed on a cloud computing platform for bridge health monitoring. The synchronization of data between the physical and virtual systems plays a crucial role in these approaches. Gao et al. (2023) proposed a communication framework that integrates edge devices and cloud servers to achieve efficient and low-latency data transmission. Similarly, Broo et al. (2022), Broo and Schooling (2021) developed a 4-layer framework comprising the physical, cyber, integration, and service layers. The physical layer includes a real bridge and sensing devices. Data collected from various sources are processed on a server. The cyber layer comprises cloud-based software tools, while the integration layer fuses data and transforms the model into visualized results. The service layer facilitates communication between systems using network protocols. However, this framework primarily focuses on the operation and maintenance phases of the bridge and does not cover other phases or provide services for case validation.

A few BDT frameworks have been proposed that are based on various types of models. Shim et al. (2019a) proposed a framework for maintaining small and medium span bridges, incorporating digital twin models built on a BIM platform, reality twin models created with drone photography and 3D laser scanning, and mechanical twin models. However, this approach divides the implementation of digital twins into building multiple models, limiting the self-updating capabilities and the ability to reflect the interactivity of diverse data and services. Integrating the digital twin into an existing BIM framework enables unified data format and information sharing (Song et al., 2023), though it was developed for operation and maintenance.

In summary, various digital twin frameworks have been proposed in bridge engineering, incorporating different elements and models. These frameworks aim to enhance the management, monitoring, and decision-making processes throughout the life cycle of bridges. However, further research is needed to develop comprehensive frameworks that cover all phases and provide robust services for validation.

A BDT framework based on the entire life cycle of bridges

A comprehensive bridge digital twin (BDT) framework is proposed to address the existing limitations, focusing on the entire life cycle of a bridge. Integrating the design approach of (Broo et al., 2022; Broo and Schooling, 2021) and technologies like BrIM, cloud computing, big data, and machine learning, the framework revolves around data, models, and service orientation.

As shown in Figure 7, the proposed framework comprises four key components: the physical system, virtual system, data system, and services system. The physical system includes the asset layer, perception layer, and communication layer. The asset layer comprises bridge components and external excitations, while the perception layer incorporates sensors to measure bridge characteristics. The communication layer facilitates local and remote data transmission. The virtual system serves as the intelligent core, comprising the operating layer, model layer, and visualization layer. The operating layer acts as the control platform, supporting cloud computing and simulation. The model layer includes geometric, analysis, and simulation models. The visualization layer presents the virtual bridge, graphical user interface (GUI) for interaction, and service program interface. The data system is crucial, encompassing the storage layer and integration layer. The storage layer manages design, construction, and O&M databases, while the integration layer handles data pre-processing, fusion, and mining. The services system is an open application platform offering services throughout the bridge's life cycle. It ensures compatibility with different phases and modularity of the BDT framework.

Figure 7.

A BDT framework based on the entire life cycle of bridges.

The proposed framework provides enhanced intelligence and functionality for bridge management and decision making processes by emphasizing data, model and service orientation and leveraging technologies such as BrIM, cloud computing, big data and machine learning. Although not yet deployed in actual bridge projects, the framework can serve as an alternative to implementing digital twins throughout the bridge lifecycle.

Discussion

This section discusses the main challenges of applying digital twins in bridge engineering, future directions, and the limitations of this study based on the previous research and analysis.

Research challenges and future directions

Non-technological perspective

The lack of standardized specifications and systematic thinking regarding the concept and implementation of digital twins in bridge engineering is a significant challenge highlighted in this research. Firstly, the definition of BDT remains ambiguous, and there is confusion with BrIM in some publications, hindering the widespread application of digital twins in bridge engineering. Despite ongoing efforts to establish a consistent understanding of BDT, a widely accepted formal definition is yet to be established. Additionally, there is a lack of established standards and protocols for twinning between physical and virtual bridges, resulting in virtual twins with varying levels of fidelity.

Secondly, while some digital twin frameworks have been proposed for specific aspects of bridge O&M, such as fatigue life prediction and proactive maintenance measures, a comprehensive framework covering the entire life cycle of bridges is yet to be presented. Implementing BDT involves integrating multidisciplinary technologies and knowledge from bridge engineering, computer science, and communication engineering, further complicating the design of a comprehensive technical architecture.

Finally, BDT currently lacks a systematic approach to service delivery. Research efforts primarily focus on geometric model generation, FEM updating, and maintenance and management of existing bridges, indicating that BDT services are predominantly targeted towards the operational and maintenance phases, with limited involvement in the design and construction phases. While BDT has the potential to be used throughout the entire life cycle of bridges, there are significant constraints when applying the BDT concept to the design and construction stages.

Therefore, it is essential to establish a clear and unified description of the definition, characteristics, components, and services of BDT, along with standardized BDT systems for researchers and practitioners. Furthermore, the design and validation of the BDT framework are crucial processes for the effective deployment of digital twins in bridge engineering, including virtual parts, data formats, and service applications. Bridging this gap will require increased resources and exploration of how digital twins can be introduced at the design and construction stages of bridges.

Technological perspective

Building virtual bridges is a significant technological challenge in implementing BDT. For existing bridge projects, generating the gDT through laser scanning or photogrammetry techniques can track the variations of the real bridge's geometric features. However, these techniques have limitations in terms of quality and efficiency, especially for complex bridges with curved decks. Regarding new bridge projects, there is a lack of research on building virtual bridges from the design phase to enable digital twin applications throughout the bridge's life cycle. Furthermore, the gDT of the bridge does not imply the realization of BDT since the mechanical information of the real bridge is not simultaneously updated.

FEM updating is a commonly used method to update the structural properties of bridges. However, the reliability of the updated FEM in fully representing the real bridge structure needs further verification and validation. Linear models, typically used in FEM updating studies, may not be applicable to nonlinear structural systems under extreme loads like earthquakes. Additionally, FEM updating is an inverse problem that involves deriving revised FEM parameters or assumptions using structural response data obtained from sensor measurements, introducing uncertainties that can affect the reliability of the revision results. Another drawback of FEM updating is its inability to update geometric properties such as section shape and area, limiting the accuracy and broader application of BDT.

The production of virtual bridges requires more attention as a critical component of BDT implementation. Research should focus on leveraging technologies like BrIM and GIS to establish virtual bridges for new bridge projects. Furthermore, new techniques and methods for gDT generation of existing bridges should be developed to achieve full automation and higher accuracy. Intelligent algorithms and updating techniques should also be continuously researched to improve the real-time and reliability of FEM updating. Combining FEM updating with geometric model generation is expected to be a prominent future trend for creating virtual bridges, compensating for the drawbacks of each method.

Data and connectivity pose another significant technical challenge for BDT. Data serves as the core element of BDT and enables the connection between the virtual bridge and the physical bridge. Point cloud data from laser scanners and image data from cameras are commonly used for geometric characterization of bridges. Structural modal, stress, and strain data acquired through sensors contribute to characterizing the structural state of bridges. However, accessing comprehensive data obtained from the physical bridge remains a key factor in the virtual bridge's ability to provide services. Challenges include processing massive point cloud data, accurately describing bridge geometry using image data, and distinguishing between changes in bridge structures and environmental conditions using sensor data. Developing intelligent and reliable techniques and methods for all-element data collection is crucial for effective BDT implementation. Integrating multi-source heterogeneous data is another pressing challenge in BDT development. While individual studies have utilized specific functions of BDT using acquired data, integrated utilization of these data sources has not been widely demonstrated. Integrating the data flow throughout the stages of bridge design, construction, operation, and maintenance is crucial for unified management and maximizing the application value of BDT.

To address these challenges, future research should focus on introducing BDT from the design stage and utilizing data acquired from the physical bridge to establish near real-time twinning with subsequent construction, operation, and maintenance stages. Achieving real-time connection and bi-directional data flow between physical and virtual bridges is key to realizing the use of digital twins throughout the bridge's entire life cycle.

Limitations

Despite conducting a comprehensive systematic literature review, this study has certain limitations. Firstly, the search for relevant documents was limited to only two literature databases, Scopus and Web of Science, potentially excluding other publications on digital twins in bridge engineering. Secondly, while strict selection criteria were applied during the screening process, the subjective nature of developing these criteria and conducting manual screening introduces the possibility of bias in the number and relevance of publications included in the review. Additionally, the categorization of research directions and analysis of digital twin applications at different stages of the bridge's life cycle rely on subjective judgment, which can impact the accuracy and reliability of the findings. Overall, these limitations may restrict the generalizability of the study, but they still offer valuable insights and reflections for practitioners working with digital twins in bridge engineering.

Conclusions

This study offers a systematic review of digital twins in bridge engineering, presenting a clear definition of BDT and summarizing its key features and creation methods. A detailed comparison between BDT and BrIM technologies is conducted, highlighting their distinctions in terms of constituent elements, characteristics, fidelity, services, and degree of development. While BrIM focuses on simulating historical bridge states, BDT enables real-time twinning to continuously represent changes throughout the bridge's life cycle.

In addition, this review highlights the research distribution of BDT, focusing on geometric model generation, FEM updating, management and maintenance. It identifies key applications of BDT at different stages of a bridge’s life cycle, emphasizing the need for further development in design and construction phases. BDT is recognized as an optimal platform for integrating various technologies such as BrIM, GIS, SHM, laser scanning, and FEM updating to create a comprehensive smart bridge throughout its entire life cycle. To facilitate industry adoption, a reference framework centered on data and models, service-oriented, and spanning the bridge's life cycle is proposed based on the findings of this review.

Finally, the challenges of using digital twins in bridge engineering are discussed, along with suggestions for future research. It is essential to establish standardized definitions, characteristics, and service functions for BDT, as well as develop a standardized system for its implementation throughout the bridge life cycle. On the other hand, the techniques, algorithms and tools for creating virtual bridges should be continuously improved and developed. The effective data collection and processing for all bridge elements play a pivotal role in ensuring the accuracy and twinning rate of virtual bridges. Furthermore, there is a growing emphasis on expanding the application of BDT to cover all stages of bridge development and maintenance.

Footnotes

Acknowledgments

The financial support is gratefully acknowledged.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is funded by the National Natural Science Foundation of China (Project No 52208150), the Natural Science Foundation of Jiangsu Province (Project No. BK20220853), and the development of bridge digital twin for operation and maintenance platform in Wuxi (H202210208).

ORCID iD

Yichen Zhu

References

Akanbi

Zhang

(2022) Framework for developing IFC-based 3D documentation from 2D bridge drawings. Journal of Computing in Civil Engineering 36(1): 04021031.

Bai

Huan

Kim

(2012) Measuring bridge construction efficiency using the wireless real-time video monitoring system. Journal of Management in Engineering 28(2): 120–126.

Baisthakur

Chakraborty

(2021) Experimental verification for load rating of steel truss bridge using an improved Hamiltonian Monte Carlo-based Bayesian model updating. Journal of Civil Structural Health Monitoring 11(4): 1093–1112.

Bello

Popescu

Blanksvärd

, et al. (2022) Framework for bridge management systems (bms) using digital twins. Lecture Notes in Civil Engineering 687–694.

Biancardo

Viscione

Oreto

, et al. (2020) BIM approach for modeling airports terminal expansion. Infrastructures 5(5): 41.

Bittencourt

Futai

da Conceição Neto

, et al. (2021) Digital transformation of bridges inspection, monitoring and maintenance processes. In: Frangopol

Yokota

(eds), Bridge Maintenance, Safety, Management, Life-Cycle Sustainability and Innovations - Proceedings of the 10th International Conference on Bridge Maintenaince, Safety and Management, IABMAS 2020. CRC Press/Balkema, 2021; 11–30.

Broo

Schooling

(2021) A framework for using data as an engineering tool for sustainable cyber-physical systems. IEEE Access 9: 22876–22882.

Broo

Bravo-Haro

Schooling

(2022) Design and implementation of a smart infrastructure digital twin. Automation in Construction 136: 104171.

Cheng

JCP

Deng

(2016) Analytical review and evaluation of civil information modeling. Automation in Construction 67: 31–47.

10.

Chipman

Costin

Yang

, et al. (2016) Bridge Information Modeling Standardization Report, Volume I – Information Exchanges. Report No. FHWA-HIF-16-011, 1 April. Federal Highway Administration. Office of Bridges and Structures.

11.

Costin

Medlock

(2021) Building information modeling for bridges and structures: outcomes and lessons learned from the steel bridge industry. Transportation Research Record 2675(11): 576–586.

12.

Dan

Ying

(2021) Digital twin system of bridges group based on machine vision fusion monitoring of bridge traffic load. IEEE Transactions on Intelligent Transportation Systems 23(11): 22190–22205.

13.

Dang

Shim

(2020) Bridge assessment for PSC girder bridge using digital twins model. Lecture Notes in Civil Engineering 54: 1241–1246.

14.

Dang

Tatipamula

Nguyen

(2022) Cloud-based digital twinning for structural health monitoring using deep learning. IEEE Transactions on Industrial Informatics 18(6): 3820–3830.

15.

Falagas

Pitsouni

Malietzis

, et al. (2008) Comparison of pubmed, scopus, web of science, and google scholar: strengths and weaknesses. The FASEB Journal 22(2): 338–342.

16.

Fanning

Clevenger

Ozbek

, et al. (2015) Implementing bim on infrastructure: comparison of two bridge construction projects. Practice Periodical on Structural Design and Construction 20(4): 04014044.

17.

Febrianto

Butler

Girolami

, et al. (2022) Digital twinning of self-sensing structures using the statistical finite element method. Data-Centric Engineering 3(3): e31.

18.

Futai

Bittencourt

Carvalho

, et al. (2022) Challenges in the application of digital transformation to inspection and maintenance of bridges. Structure and Infrastructure Engineering 18(10–11): 1581–1600.

19.

Gao

Xiong

, et al. (2023) AIoT-informed digital twin communication for bridge maintenance. Automation in Construction 150: 104835.

20.

Ghahari

Malekghaini

Ebrahimian

, et al. (2022) Bridge digital twinning using an output-only bayesian model updating method and recorded seismic measurements. Sensors 22(3): 1278.

21.

Girardet

Boton

(2021) A parametric BIM approach to foster bridge project design and analysis. Automation in Construction 126: 103679.

22.

Glaessgen

Stargel

(2012) The digital twin paradigm for future NASA and U.S. air force vehicles. In: 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, 2012.

23.

Göçer

Hua

Göçer

(2016) A BIM-GIS integrated pre-retrofit model for building data mapping. Building Simulation 9(5): 513–527.

24.

Grieves

(2014) Digital twin: manufacturing excellence through virtual factory replication. White paper 1: 1–7.

25.

Grieves

Vickers

(2017) Digital twin: mitigating unpredictable, undesirable emergent behavior in complex systems. Transdisciplinary Perspectives on Complex Systems: New Findings and Approaches 85–113.

26.

Guo

X-Y

Fang

S-E

(2023) Digital twin based lifecycle modeling and state evaluation of cable-stayed bridges. Engineering with Computers 1–15.

27.

Hallaji

Fang

Winfrey

(2022) Predictive maintenance of pumps in civil infrastructure: state-of-the-art, challenges and future directions. Automation in Construction 134: 104049.

28.

Han

Qin

, et al. (2023) Semi-automated generation of geometric digital twin for bridge based on terrestrial laser scanning data. Advances in Civil Engineering 2023.

29.

Huang

S-F

Chen

C-S

Dzeng

R-J

(2011) Design of track alignment using building information modeling. Journal of Transportation Engineering 137(11): 823–830.

30.

Isailović

Stojanovic

Trapp

, et al. (2020) Bridge damage: detection, IFC-based semantic enrichment and visualization. Automation in Construction 112: 103088.

31.

Jiang

Ding

Song

, et al. (2021a) An architecture of lifecycle fatigue management of steel bridges driven by digital twin. Structural Monitoring and Maintenance 8(2): 187–201.

32.

Jiang

Ding

Song

, et al. (2021b) Digital twin-driven framework for fatigue life prediction of steel bridges using a probabilistic multiscale model: application to segmental orthotropic steel deck specimen. Engineering Structures 241: 112461.

33.

Jiang

Broyd

, et al. (2021c) Digital twin and its implementations in the civil engineering sector. Automation in Construction 130: 103838.

34.

Jiang

Ding

Song

, et al. (2022a) Digital Twin-driven framework for fatigue lifecycle management of steel bridges. Structure and Infrastructure Engineering 1–21.

35.

Jiang

Broyd

, et al. (2022b) Underpass clearance checking in highway widening projects using digital twins. Automation in Construction 141: 104406.

36.

Jones

Snider

Nassehi

, et al. (2020) Characterising the digital twin: a systematic literature review. CIRP Journal of Manufacturing Science and Technology 29: 36–52.

37.

Kim

(2012) RFID and CCTV-based material delivery monitoring for cable-stayed bridge construction. Journal of Computing in Civil Engineering 26(2): 183–190.

38.

Kaewunruen

Sresakoolchai

, et al. (2021) Digital twin aided vulnerability assessment and risk-based maintenance planning of bridge infrastructures exposed to extreme conditions. Sustainability 13(4): 1–19.

39.

Kang

Kim

Moon

, et al. (2016) Managing construction schedule by telepresence: integration of site video feed with an active nD CAD simulation. Automation in Construction 68: 32–43.

40.

Kang

J-S

Chung

Hong

(2021) Multimedia knowledge‐based bridge health monitoring using digital twin. Multimedia Tools and Applications 80(26–27): 34609–34624.

41.

Khajavi

Motlagh

Jaribion

, et al. (2019) Digital twin: vision, benefits, boundaries, and creation for buildings. IEEE Access 7: 147406–147419.

42.

Khan

Khalid

Anjum

, et al. (2022) Tag and IoT based safety hook monitoring for prevention of falls from height. Automation in Construction 136: 104153.

43.

Lai

Chen

Hong

, et al. (2022) Framework for long-term structural health monitoring by computer vision and vibration-based model updating. Case Studies in Construction Materials 16: e01020.

44.

Lee

Shim

, et al. (2012) Bridge information models for construction of a concrete box-girder bridge. Structure and Infrastructure Engineering 8(7): 687–703.

45.

Lee

Lapira

Bagheri

, et al. (2013) Recent advances and trends in predictive manufacturing systems in big data environment. Manufacturing Letters 1(1): 38–41.

46.

Chan

NKY

Huang

, et al. (2012) Virtual prototyping for planning bridge construction. Automation in Construction 27: 1–10.

47.

Xue

, et al. (2018) An Internet of Things-enabled BIM platform for on-site assembly services in prefabricated construction. Automation in Construction 89: 146–161.

48.

Xiao

Guo

, et al. (2022) A bim based approach for structural health monitoring of bridges. KSCE Journal of Civil Engineering 26(1): 155–165.

49.

Liu

Fang

Dong

, et al. (2021) Review of digital twin about concepts, technologies, and industrial applications. Journal of Manufacturing Systems 58: 346–361.

50.

Brilakis

(2019) Digital twinning of existing reinforced concrete bridges from labelled point clusters. Automation in Construction 105: 102837.

51.

Guo

, et al. (2020) An automated target-oriented scanning system for infrastructure applications. In: El Asmar

Tang

Grau

(eds), Construction Research Congress 2020: Computer Applications - Selected Papers from the Construction Research Congress 2020, Tempe, Arizona, 8–10 March 2020, 2020; 457–467.

52.

Marra

Trizio

Fabbrocino

, et al. (2021) Digital models for e-conservation: the HBrIM of a bridge along the Aterno river. SCIRES-IT - SCIentific RESearch and Information Technology 11(2): 83–96.

53.

Marzouk

Hisham

(2014) Implementing earned value management using bridge information modeling. KSCE Journal of Civil Engineering 18(5): 1302–1313.

54.

McGuire

Atadero

Clevenger

, et al. (2016) Bridge information modeling for inspection and evaluation. Journal of Bridge Engineering 21(4): 04015076.

55.

Meixedo

Santos

Ribeiro

, et al. (2022a) Online unsupervised detection of structural changes using train-induced dynamic responses. Mechanical Systems and Signal Processing 165: 108268.

56.

Meixedo

Ribeiro

Santos

, et al. (2022b) Real-time unsupervised detection of early damage in railway bridges using traffic-induced responses. Structural Integrity 21: 117–142.

57.

Mohammadi

Rashidi

Mousavi

, et al. (2021) Quality evaluation of digital twins generated based on uav photogrammetry and tls: bridge case study. Remote Sensing 13(17): 3499.

58.

Morgenthal

Hallermann

Kersten

, et al. (2019) Framework for automated UAS-based structural condition assessment of bridges. Automation in Construction 97: 77–95.

59.

Negri

Fumagalli

Macchi

(2017) A review of the roles of digital twin in CPS-based production systems. Procedia Manufacturing 11 In: 27th International Conference on Flexible Automation and Intelligent Manufacturing, FAIM2017, Modena, Italy, 27-30 June 2017, 939–948.

60.

Nguyen

Jin

, et al. (2021) BIM-based mixed-reality application for bridge inspection and maintenance. Construction Innovation 22(3): 487–503.

61.

Omer

Margetts

Hadi Mosleh

, et al. (2019) Use of gaming technology to bring bridge inspection to the office. Structure and Infrastructure Engineering 15(10): 1292–1307.

62.

Omer

Margetts

Mosleh

, et al. (2021) Inspection of concrete bridge structures: case study comparing conventional techniques with a virtual reality approach. Journal of Bridge Engineering 26(10): 05021010.

63.

Opoku

DGJ

Perera

Osei Kyei

, et al. (2021) Digital twin application in the construction industry: a literature review. Journal of Building Engineering 40: 102726.

64.

Page

McKenzie

Bossuyt

, et al. (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. International Journal of Surgery 88: 105906.

65.

Pantoja-Rosero

Saloustros

Achanta

, et al. (2023) Image-based geometric digital twinning for stone masonry elements. Automation in Construction 145: 104632.

66.

Pregnolato

Gunner

Voyagaki

, et al. (2022) Towards civil engineering 4.0: concept, workflow and application of digital twins for existing infrastructure. Automation in Construction 141: 104421.

67.

Tao

(2018) Digital twin and big data towards smart manufacturing and industry 4.0: 360 degree comparison. IEEE Access 6: 3585–3593.

68.

Rashidi

Karan

(2018) Video to BrIM: automated 3D as-built documentation of bridges. Journal of Performance of Constructed Facilities 32(3): 04018026.

69.

Rathore

Shah

Shukla

, et al. (2021) The role of ai, machine learning, and big data in digital twinning: a systematic literature review, challenges, and opportunities. IEEE Access 9: 32030–32052.

70.

Sacks

Kedar

Borrmann

, et al. (2018) Seebridge as next generation bridge inspection: overview, information delivery manual and model view definition. Automation in Construction 90: 134–145.

71.

Sánchez-Rodríguez

Esser

Abualdenien

, et al. (2020) From point cloud to IFC: a masonry arch bridge case study. In: Hartmann

Ungureanu

L-C

(eds), EG-ICE 2020 Workshop on Intelligent Computing in Engineering, 422–431.

72.

Shao

Zhou

Deng

, et al. (2020) Experiment of structural geometric morphology monitoring for bridges using holographic visual sensor. Sensors 20(4): 1187.

73.

Shim

Dang

Lon

, et al. (2019a) Development of a bridge maintenance system for prestressed concrete bridges using 3D digital twin model. Structure and Infrastructure Engineering 15(10): 1319–1332.

74.

Shim

Kang

Dang

(2019b) Digital twin models for maintenance of cable-supported bridges. In: Viggiani

GMB

DeJong

Schooling

(eds), International Conference on Smart Infrastructure and Construction 2019, ICSIC 2019: Driving Data-Informed Decision-Making, ICE Publishing, 2019; 737–742.

75.

Shu

Zandi

Topac

, et al. (2019) Automated generation of FE model for digital twin of concrete structures from segmented 3D point cloud. In: Kopsaftopoulos

Chang

F-K

Guemes

(eds), Structural Health Monitoring 2019: Enabling Intelligent Life-Cycle Health Management for Industry Internet of Things (IIOT) - Proceedings of the 12th International Workshop on Structural Health Monitoring, DEStech Publications Inc, 2019; 428–434.

76.

Song

Yang

, et al. (2023) Digital twin enhanced BIM to shape full life cycle digital transformation for bridge engineering. Automation in Construction 147: 104736.

77.

Tao

Cheng

, et al. (2018) Digital twin-driven product design, manufacturing and service with big data. The International Journal of Advanced Manufacturing Technology 94(9): 3563–3576.

78.

Tao

Zhang

Liu

, et al. (2019a) Digital twin in industry: state-of-the-art. IEEE Transactions on Industrial Informatics 15(4): 2405–2415.

79.

Tao

Sui

Liu

, et al. (2019b) Digital twin-driven product design framework. International Journal of Production Research 57(12): 3935–3953.

80.

Van Nimmen

Van Hauwermeiren

van den Broeck

(2021) Eeklo footbridge: benchmark dataset on pedestrian-induced vibrations. Journal of Bridge Engineering 26(7): 05021007.

81.

Vivi

Parlikad

Woodall

, et al. (2019) Developing a dynamic digital twin at a building level: using cambridge campus as case study. In: Viggiani

GMB

DeJong

Schooling

(eds), International Conference on Smart Infrastructure and Construction 2019, ICSIC 2019: Driving Data-Informed Decision-Making, ICE Publishing, 2019; 67–75.

82.

Yang

Cheng

JCP

Wang

(2020) Semi-automated generation of parametric BIM for steel structures based on terrestrial laser scanning data. Automation in Construction 112: 103037.

83.

Yang

Castillo

Zou

, et al. (2022) Automated semantic segmentation of bridge components from large-scale point clouds using a weighted superpoint graph. Automation in Construction 142: 104519.

84.

Butler

Calka

, et al. (2019) A digital twin of bridges for structural health monitoring. In: Kopsaftopoulos

Chang

F-K

Guemes

85.

Kuok

Butler

, et al. (2020a) Implementing bridge model updating for operation and maintenance purposes: examination based on UK practitioners’ views. Structure and Infrastructure Engineering 18(12): 1638–1657.

86.

Lai

Bartoli

, et al. (2020b) Technology for condition and performance evaluation of highway bridges. Journal of Civil Structural Health Monitoring 10(4): 573–594.

87.

Yoon

Lee

Kye

, et al. (2022) Seismic fragility analysis of deteriorated bridge structures employing a UAV inspection-based updated digital twin. Structural and Multidisciplinary Optimization 65(12): 346.

88.

(2022) Digital twin-based structure health hybrid monitoring and fatigue evaluation of orthotropic steel deck in cable-stayed bridge. Structural Control and Health Monitoring 29(8): e2976.

89.

Zhao

Tan

OBrien

, et al. (2022) Developing digital twins to characterize bridge behavior using measurements taken under random traffic. Journal of Bridge Engineering 27(1): 04021101.

90.

Zhou

Xiao

, et al. (2022) An example of digital twins for bridge monitoring and maintenance: preliminary results. Lecture Notes in Civil Engineering 1134–1143.

91.

Zou

Kiviniemi

Jones

, et al. (2019) Risk information management for bridges by integrating risk breakdown structure into 3D/4D BIM. KSCE Journal of Civil Engineering 23(2): 1–14.