Toward Sustainable Development: Exploring the Value and Benefits of Digital Twins

Abstract

The complexity and number of data streams generated by internal processes exceed the capabilities of most current simulation environments. Consequently, there is a need for the development of more advanced solutions that can handle any number of simultaneous simulations. One of the most promising ideas to address these and other challenges is the concept of a Digital Twin (DT), which refers to a digital representation or a virtual model designed to accurately reflect an intended or actual physical product, system, or process (i.e., a physical twin). As a Digital Twin spans the life-cycle of its physical twin, its development and application can bring considerable benefits to organizations seeking to improve existing processes as well as implement new ones. However, few studies have comprehensively examined the value and benefits of Digital Twins. To fill this gap, this study aims to provide a better understanding of this technology by reviewing the contemporary literature, with a particular focus on the documented case studies, as well as reported business and industrial deployments. The results obtained show that Digital Twins have proven beneficial for maintenance, cost reduction, optimization, simulation performance, monitoring, product life-cycle understanding, assessment validation, performance evaluation, product design, and safety and risk mitigation. In addition, when considering the human factor, DTs can facilitate education and training, team collaboration, and decision making. Undeniably, Digital Twins are a game changer for safer, faster, and more sustainable development.

1. Introduction

Sustainable development is development that “meets the needs of the present generation without compromising the ability of future generations to meet their own needs” [1]. In general, this issue has received considerable attention from businesses, governments, and researchers over the last decades [2,3,4]. Not surprisingly, due to global warming and its effects [5], there is a clear and growing interest in developing and implementing low-impact solutions [6]. Among many key drivers for sustainability, an innovation is one of today’s most desired technological traits to create more efficient and more sustainable products and services [7,8]. According to Prudy et al. [9], Digital Twins are changing the face and pace of innovation, while the implications are profound, making an innovation faster, cheaper, and more radical.

According to Fortune Business Insights, the global Digital Twin market size reached a value of US$6.75 billion in 2021 and is projected to reach US$96.49 billion by 2029, exhibiting a compound annual growth rate (CAGR) of 40.6% during the forecast period [10]. The world’s largest tech companies, including Alphabet [11], Amazon [12], and Microsoft [13], offer development tools that enable the creation of Digital Twins of real-world systems in areas such as construction [14], healthcare [15], manufacturing [16], supply chain [17], and retail [18], to name a few. Moreover, Digital Twins and closely related areas have piqued the interest of researchers from over 158 countries (The number of countries has been estimated based on the results from the query “digital twin” execution on the Scopus database (30 March 2023)), spanning a wide range of fields, including engineering, computer science, physics, energy, biochemistry, decision sciences, and business management. Despite the numerous research questions raised and answered, a fundamental question still remains: what value and benefits can a Digital Twin deployment bring to the organization?

In this paper, we aim to explore the value of a Digital Twin and identify its benefits, regardless of its domain of application. Thus, our study is qualitative in nature, providing an appropriate way to analyze a relatively under-examined topic. In other words, our study relies on exploring the state-of-the-art literature focused on Digital Twin development and implementation in any area of business or industry. In this sense, this paper contributes to the limited research in the Digital Twin landscape by identifying and articulating the story of its value and benefits, and informing interested parties of possible outcomes resulting from its adoption.

The rest of the paper is structured as follows. In Section 2, we outline the theoretical and technological background that underpins the Digital Twin concept and architecture. In Section 3, we present the research methodology we applied to achieve our research objectives. In Section 4, we provide the results we obtained, followed by a discussion of threats to validity, as well as contributions, limitations, and future research directions in Section 5.

2. Background

It is widely accepted that the concept of a “digital twin” originated from the National Aeronautics and Space Administration (NASA) in the 1960s, where it was referred to as a “living model” of the Apollo mission [19]. The term itself was first used by John Vickers in 2002 [20]. According to VanDerHorn and Mahadevan, a Digital Twin (DT) is defined as “a virtual representation of a physical system (and its associated environment and processes) that is updated through the exchange of information between the physical and virtual systems” [21].

Lueth [22] classified Digital Twin use cases along three prevailing dimensions:

• The hierarchical level at which the Digital Twin is deployed includes six levels (Informational, Component, Product, Process, System, and Multi-system).
• The life-cycle phase in which the Digital Twin is deployed involves six phases (Design, Build, Operate, Maintain, Optimize, and Decommission).
• The use of the Digital Twin refers to seven uses (Digitize, Visualize, Simulate, Emulate, Extract, Orchestrate, and Predict).

In total, 252 potential use cases are defined by combining the seven most common uses, six hierarchical levels, and six phases. It seems quite evident that researchers and practitioners have so far have documented and classified a plethora of use cases of the Digital Twin in many different sectors, introducing different sets of criteria, ranging from the general to the very technical [23].

In general, the architecture of a Digital Twin in terms of its modeling can be divided into three main layers [24,25]:

• The physical layer, which encompasses the physical entities of the system, environment, and processes;
• The virtual layer, which involves virtual models of the physical entities at different levels of abstraction;
• The interconnection layer, which serves as the bidirectional data proxy between the physical and virtual layers.

From the perspective of a solution vendor, the first physical reality layer is independent and is the object of observation, analysis, and understanding; while the remaining two dependent layers are the object of modeling, implementing, and testing. Four core technologies typically employed in modeling and developing a DT are Artificial Intelligence (AI), Big Data (BD), Internet-of-Things (IoT), and mobile/wireless networks [26].

AI aims to empower a Digital Twin with capabilities of perceiving, synthesizing, and inferring information to learn and adapt to new situations and circumstances [27]. Some researchers and practitioners argue that AI is an imperative component for some DT applications, such as autonomous driving systems [28] and additive manufacturing [29]. In particular, machine learning algorithms are commonly used to explore data for hidden patterns [30], detect anomalies [31], make predictions [32], or construct classifiers [33].

Big Data (BD) spans over the “high” seven Vs—namely, Volume, Velocity, Variety, Veracity, Variability, Visualization, and Value [34]. In short, Big Data refers to relatively larger, more complex data repositories [35], especially from new data sources [36]. For Perry [37], the concept Big Data really means “harvesting meaning from data”, which “is coming in faster, from more sources, and in more varied formats than ever before” [38]. In a practical sense [39], Big Data technology refers to a software utility designed to “analyze process and extract information from extremely complex and large data sets which traditional data processing software could never handle” [40]. The modern big data solutions available in the market are [41] Apache Hadoop [42], Apache Spark [43], and Hortonworks Data Platform (HDP) [44].

The term Internet of Things (IoT) is understood as “the collective network of connected devices and the technology that facilitates communication between devices and the cloud, as well as between the devices themselves” [45]. In general, IoT capacitates the Digital Twin model to support new intelligent services to connect and interact with physical objects [46]. The realms of IoT and DTs overlap when it comes to describing, discovering, and accessing resources [47]. This is conceptualized on five layers, starting from physical space, communication network, virtual space, data analysis and virtualization, and ending with the application layer [48]. In the case of the first layer, the naming of objects is typically associated with the word “smart” when IoT comes into action; hence, numerous studies have already discussed smart buildings [49], smart enterprises [50], smart factories [51,52], smart farming [53], smart grid [54], smart health [55], and smart transportation [56], to name a few.

In recent years, significant advancements in mobile and wireless technologies have enabled many applications beyond traditional voice and video calls on mobile phones [57]. Digital Twin solutions are no exception to this trend. One of the main assumptions underlying the design of Digital Twins is reliability [58], which implies that twins are synchronized in real-time with their physical counterparts by receiving actual data with low latency. Data from physical entities are collected through Data Acquisition Systems (DAC) deployment [59] equipped with sensors, measurement devices, and computers [60].

The concept of a Digital Twin is not merely a passing trend from a business perspective. Even industries such as aviation and wind turbine manufacturing, which traditionally have little to do with digitization, have embraced the latest technological advancements and adapted their existing systems and procedures to incorporate Digital Twins. Recently, the term Digital Twin is often used to refer to the latest wave of computer modeling and simulation, which aim to develop digital representations of real-world objects, systems, or processes. Ultimately, the application of Digital Twins is expected to lead to significant time and cost savings [61].

3. Methodology

In our study, we put forward the following research question: what is the value and benefits of deploying Digital Twins across various applications? Given the nature of this research, it makes sense to adopt a qualitative approach using a literature review, which by definition is a structured, comprehensive review with an explicit methodology that is focused on finding the key evidence needed to answer the research question [62]. In general, according to Knopf [63], a literature review consists of two parts. First, it should be a concise summary of the findings or claims that are the result of previous research efforts on a topic. Second, it should provide a conclusion about how accurate and complete that knowledge is. In this sense, a literature review process “involves researching, reading, analyzing, evaluating, and summarizing scholarly literature (typically journals and articles) about a specific topic” [64].

Our research inquiry begins by specifying data sources. For the primary data source, we used Google Scholar (available at https://scholar.google.com/), which is the largest academic database with over 390 million indexed records [65]. As a secondary data source, we used Scopus (available at https://www.scopus.com/), an Elsevier abstract and citation multidisciplinary database of peer-reviewed literature, which includes scientific journals, books, and conference proceedings [66]. Note that both Google Scholar and Scopus are legitimate, reliable, and up-to-date sources of information used by most researchers around the world [67,68].

We developed search queries by combining keywords related to Digital Twin research, such as advantage, benefit, and value. Our objective was to analyze and summarize the existing literature comprehensively by discussing the background of the topic. We conducted a narrative literature review as the first step in our study [69]. Please note that for the Scopus database, we only searched for titles of reviewed articles and conference papers published solely in English.

The second step involved searching, analyzing, and documenting existing Digital Twin deployments across all areas of business or industry. We used a combination of keywords such as application, deployment, design, implementation, and solution. More specifically, we employed an exploratory approach to investigate the benefits reported after Digital Twin deployments. We excluded studies that only discussed promises or expectations. If insufficient information was found in the collected scholarly literature, we also used Google Search Engine (available at https://www.google.com/), as a supplementary data source.

Eventually, having collected necessary data, we performed manual coding, following the instructions authored by Hacking [70]. By definition, in qualitative research, a code is usually a word or short phrase that semantically assigns a summative attribute for a portion of language-based data. Therefore, the application of manual coding involves detailed reading and manually developing and assigning codes.

In this study, we adopted the inductive coding approach, which means building a list of codes from scratch based on the collected data. In this extent, the following four steps were undertaken. Firstly, one of the researchers read all the papers and extracted all the text fragments related to the research question. Secondly, by making use of the keywords and their synonyms, the contextualized meaning was individually classified to one word, or alternatively to a string, consisting of two or three words. Thirdly, the second researcher manually verified the coding and, if necessary, extended the list of codes. In the last step, the list was the subject of shared processing with the aim of consensual understanding and final refinement. The outcome of the manual coding is provided in Appendix A.

4. Benefits of Digital Twin Deployments

To summarize our results, we used the cloud of words technique, which is a free and online tool (worditout.com). This technique generates a visual representation of word frequency by extracting words from the source text and emphasizing those that appear more frequently. While there were 103 codes found in the original source, we set up the minimum frequency on two. Next, after careful analysis, we excluded four words (commissioning, failure, problem, and risk) due to their negative connotation, as well eight common words (analysis, cost, costs, data, interaction, model, sales, and time). Thus, the remaining set, as an input for visualization, includes 40 words. Figure 1 depicts the cloud of codes assigned, expected, and reported from Digital Twin deployment that were anticipated from our quantitative analysis.

As one can notice, the benefits of deploying a Digital Twin are multifaceted (see Figure 2). The most frequent and common benefits include the following:

• Maintenance (22.4%)—related to both operational and predictive analysis, enabling stakeholders to better understand the physical systems and actual processes;
• Cost reduction (16.3%)—refers to the better understanding of the product life-cycle loop, thus recognizing the points of materials and work overhead;
• Optimization (12.2%)—due to the discovery of the best solutions, incorporating the cost-effective resources while maximizing business value;
• Simulation performance (10.2%)—having the capability of handling multiple simultaneous and heterogeneous simulations and aims to model the processes’ evolution by testing different settings of their key behaviors, characteristics, or relationships;
• Monitoring (8.2%)—refers to the regular surveillance over digitized assets, processes, and services over a period of time;
• Product life-cycle understanding (6.1%)—refers to the handling of a good as it moves through the various stages of its lifespan, bringing tangible effects of reducing system load, configuring and planning assets, and checking product feasibility;
• Assessment validation (6.1%)—concerns the quality review of the assessment process, including checking that the assessment tool produces consistent results;
• Performance evaluation (6.1%)—through analyzing, developing, and categorizing a set of alternative scenarios and models before their implementation within the physical system;
• Product design (6.1%)—enabling designers to develop and test more product variations with fewer resources engaged;
• Safety and risk mitigation (6.1%)—relates to the detection, assessment, understanding, and prevention of adverse events, side effects, or any other health-related issues.

Last but not least, other benefits of a Digital Twin deployment pertained to innovation fostering [71], system security [72], and multidomain collaboration [73]. However, it seems that the most intriguing side of the Digital Twin implementations currently relates to the healthcare pathway involving human organs [74], such as the heart [75], liver [76], or lung [77]. In addition, Digital Twins can provide a realistic and safe environment for healthcare professionals to improve their skills by practicing complex medical procedures [78,79]. Thus, the implementation of Digital Twin technology in healthcare has the potential to greatly improve patient outcomes and support patient safety [80].

Note that while there are many adopters of digital life models, there is not yet a widely accepted consensus on the value and benefits of the Digital Twin as its first solutions are deployed and are still expanding across many industries on a global scale. Therefore, the calculated percentages should be treated with considerable caution. Nevertheless, we believe that the results obtained can provide a valuable foothold for both the academic and business communities by providing a broad view of the value and benefits of implementing Digital Twins.

5. Discussion

At present, we are witnessing the Fourth Industrial Revolution [81], and time is driven by digital transformation in the service and manufacturing sectors. Almost every industry, if not all, is currently subject to these changes—which are characterized by interconnectivity via digital tools [82,83]—that promise to deliver affordable, smart, scalable, and reliable capabilities beyond physical limitations [84]. Future generations of computer systems (FGCs), particularly Digital Twins solutions, are expected to provide support not only to their users but also to their physical counterparts in economic, operational, and strategic domains. Clearly, the FGCs are not off-the-shelf technological products; as such, a variety of unprecedented challenges and issues have already been identified [85,86,87], both in terms of their development and deployment. This makes them one of the most complex and demanding endeavors to undertake. Therefore, the expected benefits should be treated with caution and considered over the long term.

Collected information suggests that different technologies used to design and develop Digital Twin solutions have been implemented and integrated over time. There is an obvious role for hardware infrastructure, which is a major requirement for efficient data processing units as well as for the system as a whole. Another key component underpinning Digital Twin solutions is a digital definition of its counterpart, along with an attached information model. The last but not least component is a middleware that links physical and digital entities and exchanges data between them. However, combining all these components into one fully operational system is a complex and challenging process that carries additional risks to a deployment effort [88,89,90]. Thus, in addition to considering the expected benefits of a DT deployment, it is also important to identify and assess potential threats and vulnerabilities to which the system may be exposed.

Our findings are consistent with previous research that has explored the topic of Digital Twins. More generally, the domain of its applications spans many industries, including automotive, aerospace, construction, agriculture, mining, utilities, retail, healthcare, military, natural resources, and public safety sectors [91]. Secondly, since a DT is a theoretical notion and does not exist as such, and takes advantage of four technology pillars—Internet-of-Things [92], Cloud Computing [93], Artificial Intelligence [94], and Extended Reality (XR) [95]—it is able to deliver an added value, originating from the integration of the above pillars and their real-time collaboration.

While the adoption of Digital Twin technology has grown significantly in recent years, with many successful application scenarios reported to date, several challenges and issues remain to be addressed. Tahmasebinia et al. [96] identified such challenges as cost control, data processing, security, and privacy, along with seamless integration of the physical and digital worlds, and integration with intelligent solutions. In the construction domain, Madubuike et al. [97] claim that the process of deploying effective DT involves highly technical and complex steps. Digital Twin technology requires collecting and utilizing large amounts of data from multiple endpoints, each a potential vulnerability. In addition, employees may show a lack of interest due to concerns about job security, which can hinder adoption. Another significant challenge is the difficulty of integrating the various stakeholders in the construction industry to work as one team.

While Tao et al. [98] indicate that Digital Twin (DT) modeling involves physical modeling, virtual modeling, connection modeling, data modeling, and service modeling, on the other hand, such an approach implies several challenges. First, the digital model requires managing a large volume of data, including physical data, virtual data, and their integrated fusion data. As a result, data preprocessing is essential, providing data cleaning, data conversion, and data filtering. Second, all components of the digital twin must interact and collaborate to solve complex tasks. DTs have three types of interaction and collaboration: physical–physical, virtual–virtual, and virtual–physical. Note that physical–physical interaction and collaboration allows multiple physical entities to communicate, coordinate, and work together to accomplish complex tasks that a single entity cannot accomplish alone. Finally, the remaining issues relate to service encapsulation [99], service matching [100] and discovery [101], quality of service (QoS) modeling and evaluation [102,103], service optimization and integration [104,105], and fault tolerance management.

Having said that, the question that naturally arises is as follows: how do you address all these challenges and issues? Obviously, there is no single valid answer. However, considering Deponti’s view that “Digital Twin is not about technology, it’s about people. People and processes to be exact.” [106], there are many proven and well-documented methods and solutions for modeling interactions between people and processes as well as among them. Nevertheless, there is no one-size-fits-all approach that works best in all cases. Therefore, each case should be considered individually, taking into account the skills and resources [107,108] available on the one hand and the requirements and constraints [109,110] on the other.

This study was set up to investigate the benefits of Digital Twin deployments. A qualitative exploratory approach was undertaken to address this research inquiry. In our opinion, the obtained results present some interesting findings that are of value to both researchers and practitioners. Nevertheless, further research is needed to determine not only the benefits but also the challenges and limitations associated with DT deployments. One limitation of the study is that the targeted literature review only included title keyword searches of the Scopus database and only included articles and conference proceedings published in English. Therefore, future research should also incorporate more case studies and technical reports that document the deployment of DTs in real-world settings.

6. Conclusions

The deployment of Digital Twins is able to create added value by providing the capacity to analyze and evaluate both existing and under-development physical products and systems. Depending on the particular industry in question, DT models bring considerable benefits in any number of ways. However, deploying a DT also means understanding the physical products and systems and discovering their simultaneous and varied uses. Thus, one should carefully balance out the expectations and deliverables, followed by a comprehensive cost–benefit analysis. Nevertheless, while there are a number of different approaches that are possible to be used to identify and quantify the costs in an accurate and reliable way, we are not yet able to fully monetize the effects. In other words, the reported pieces of evidence so far mostly rely on qualitative judgment instead of solid numbers. Moreover, some prior studies have some drawbacks due to the lack of supporting evidence, or issues related to sampling and its representativeness, affecting the generalizability of the findings.

Despite the aforementioned weaknesses, Digital Twins could take many forms and pick up many uses, delivering value to users every single day. For instance, Google Maps, a free web mapping service with over 1 billion monthly active users [111], spanning across 220 countries and covering 24 thousand cities and towns, provides detailed information about geographical regions and sites. Here, a question arises: what will come next? Since the democratization of artificial intelligence (AI) has put advanced technologies such as ChatGPT [112] into the hands of users without specialized or even technical knowledge, the road towards Intelligent Digital Twins (IDTs) has become a reality. We believe that this shift will further empower innovators, researchers, and industry leaders from various domains to accelerate the adoption and implementation of IDTs in a wide range of everyday applications, making a significant contribution to sustainable development goals.