Digital Twins in Industry 4.0

1. Introduction

Since Grieves [1] first described the concept of digital counterparts of physical systems, which has come to be known as digital twins, digital twin technology (DT) and its related cyber–physical system (CPS) platforms have become more advanced and widespread, driving the development of Industry 4.0 as a result. Digital technologies have advanced in the areas of data storage, networks, and processing, enabling previously impossible applications. DT is used in product design and virtual prototyping, predictive maintenance, process planning, and real-time optimization. The potential benefits it offers, however, have yet to be uncovered across a product lifecycle, which includes design, engineering, manufacturing, sales, and services. Most DT applications can be found within specific sectors along certain parts of a product lifecycle. Digital twins are used in many sectors to facilitate design and to optimize production, supply chain management, control systems, and virtual/augmented reality applications. The full potential for digital twin technologies, however, is only just emerging.

In many industries, DT creates broader, more extensive sets of data across an entire product lifecycle. Cold chain management is illustrative of how integrated systems have become [2,3]. Transporting perishable goods under special conditions from production to consumers requires precision sensors throughout a long journey of traceability on common infrastructure. Sensor-laden vehicles and built environments collect data across extensive swaths of the real world, e.g., smart cities [4]. These are just a few examples of how digital technologies are expanding their reach across multiple industries and CPS platforms.

The aim of this Special Issue, therefore, is to identify potential multi-industry applications of digital twins in Industry 4.0. The articles in this Special Issue highlight aspects of innovation in digital twin architecture and sector-specific applications, including (but not limited to) the construction, education, energy, IT, manufacturing, and maritime sectors. The contributions included in this Special Issue provide a glimpse of the possibilities unfolding in DT and how the transformations that are driven by DT are only just beginning.

The following section provides an overview of the six articles included in this Special Issue. This introductory article to this Special Issue concludes with a discussion on the direction of future research based on the themes that emerge from the articles.

2. Articles in this Special Issue

DT advancements occur in the different components that form the digital infrastructure that DT is built upon—sensors, networking, storage, control, and processing. The network-connected Cloud enables the deployment of these DT applications with reduced risks, greater cost savings, and faster deployment [5]. Moreover, the increased technological capabilities of DT amplify the possibilities and challenges of modeling and designing DT systems [6]. These new applications and analytical models enable workers to diagnose more complex problems and implement more effective and efficient solutions in real time, in addition to enabling decision makers to make more complex decisions with greater precision and speed. Considering these aspects of DT, the articles in this Special Issue examine the different components of DT across the many levels of digital architecture, through various business models, and from different perspectives. Most of the current body of research heavily focuses on the application of DT to processing, but other studies are presently exploring modeling and sensing as newer technologies enable greater capabilities for different aspects of DT architecture (

Table 1. Summary of different components of DT architecture discussed in the articles in this Special Issue.

DT Application Sectors	Different Components of DT Architecture
	Modeling	Processing	Sensing
Agriculture
Biological manufacturing
Construction
Education
Energy management
Marine industry
Mechanical manufacturing
Semiconductor manufacturing
University research lab

Strongly utilized; utilized.

).

Comparing different sub-sector cases within the marine industry, Lv et al. (Contribution 1) examine the way in which to handle increased complexity through DT to identify the challenges and prospects of an integrated framework. DT enables greater predictability in modeling event processing, monitoring equipment status, and detecting environmental conditions for different purposes based on the sub-sector in question. The framework provided considers how to incorporate materials, goods, and logistics in the marine industry to improve exploration, exploitation, transportation, and processing—while simultaneously mitigating environmental impacts—along the life cycle of the entire industry and across different sub-sectors.

In the construction sector, real-time data are needed to match the progress of construction projects that frequently change the physical environment. Incorporating unmanned ground vehicles (UGVs) with advanced sensor data, including LIDAR, RADAR, and GPS, Valenzuela et al. (Contribution 2) deployed a mobile application with real-time DT geolocation mapping capabilities, with the potential for ensuring operational efficiency and safety at multiple remote building sites. UGVs enable data collection from areas that may be difficult and hazardous to access by a human worker, enhancing functionality and safety in this regard. Moreover, data processing can be handled in a centralized location where big data collected throughout a company’s locations can provide greater insights for business intelligence in the sector.

Applying DT to education, Lee et al. (Contribution 3) consider how DT can personalize and facilitate mathematics learning to improve instruction quality by going beyond measuring testing outcomes to include measurements of student interest and engagement levels. By surveying students’ perceptions while playing mathematics-based games, Lee et al. (Contribution 3) found that the digital twin gamification of math lessons could increase effectiveness and student engagement.

By considering how DT can transform agriculture, Wang (Contribution 4) reviews the state of the art in order to develop an agenda for future research. The study highlights the need to evaluate the level of data integration, the readiness of technology levels, and the changing roles of DT throughout a product lifecycle across different levels in the agricultural economy, i.e., unit, system, and system of systems levels. Wang (Contribution 4) found that most of the studies focus on monitoring and prediction components of the DT system as the most prevalent aspects, with application at the unit and system of systems being less pervasive.

Other studies focus on components of DT architecture. So et al. (Contribution 5) developed a novel hybrid tree-based ensemble learning model (HYREM) for solar irradiance forecasting using traditional climate data and gradient boosting algorithms to apply to renewable energy and eco-friendly transportation systems. The management of solar power generation and transmission depends on various issues, including weather patterns and equipment status. Thus, renewable energy management requires reliable forecasting of complex climate and operation status variables. So et al. (Contribution 5) improved upon existing prediction models by reducing errors between 44.2% and 80.1% among several metrics in the experimental results.

In another study, Kenett (Contribution 6) examines the use of emulators to optimize the performance and robustness of digital twin systems. By applying advancements in digital technologies to manufacturing, digital twin platforms provide greater capabilities for the provision of monitoring and diagnostic resources through emulators. Using these emulators, manufacturing processes become faster and more efficient, and the emulators used also enable the optimization of manufacturing system designs. Kenett (Contribution 6) demonstrates the emulator concept within the context of mechanical and biological systems processes.

Fett et al. (Contribution 7) examined the process of CPS implementation in a German university research lab to provide a common procedure for deploying middle-layer IT architecture in RAMI 4.0 (Reference Architectural Model Industry 4.0). As decision making on production processes has had greater demands on real-time responsiveness, DT in production processes provides greater capabilities for capturing and analyzing data and enables dynamic decisions to be made on the fly.

By examining semiconductor processing, Araque et al. (Contribution 8) developed a multivariable temperature control system. In their study, Araque et al. consider various aspects of the elements of physical-based, reduced-order DT control systems. While this study focused on the limits of the temperature of specific materials, new sensors will enable the modeling and design of other material conditions, as well as their related physical properties.

3. Discussion

The possibilities of DT explored in this Special Issue demonstrate new potential applications in the construction, education, energy, IT, manufacturing, and maritime sectors. While DT has incorporated many advances in digital technologies, the evolution of DT leverages innovations in sensor data, processing, and modeling—and especially in CPS. Moreover, DT transformation in Industry 4.0 has only just begun [7]. The potential for DT to expand its reach in applications across sectors will depend on the types of data, model complexity, and subsequent storage and processing power needed for more complex, automated, remote functionalities and responses in real time. For DT to reach its potential in each of these sectors, it is necessary to conduct more research in several different branches of engineering and science.

Advancements in DT will continue to provide novel applications built on new analytical models that involve greater complexity as a result of a broader scope including the integration of DT components across multiple sub-sectors—and possibly across sectors in the future [8,9]. Big data issues will continue to emerge as obstacles for DT implementation, given that big data are more sensitive to small errors as they accumulate through processing, which will require greater sensor precision. As DT is increasingly applied to industries such as smart health and personalized medicine [10,11], new technologies will continue to emerge, requiring new DT applications. As sensors develop, data will enable even greater transformations [12]. Increased data collection, however, is expected to impose greater demands on data storage and processing. Furthermore, as technologies develop, dynamic DT architecture [13] is expected to expand the capabilities and the demands needed to provide them.

Cloud architecture enables digital twins to handle big data projects, but this raises the issues of security and privacy when networks are shared. Moreover, as DT technologies are more widely applied to human–machine interactions and to the public sector [11,14,15,16], a greater emphasis on system-of-systems-level research will be required to ensure that broader environmental, social, and governance (ESG) frameworks are considered.