A state-of-the-art survey of Digital Twin: techniques, engineering product lifecycle management and business innovation perspectives

With the rapid advancement of cyber-physical systems, Digital Twin (DT) is gaining ever-increasing attention owing to its great capabilities to realize Industry 4.0. Enterprises from different fields are taking advantage of its ability to simulate real-time working conditions and perform intelligent decision-making, where a cost-effective solution can be readily delivered to meet individual stakeholder demands. As a hot topic, many approaches have been designed and implemented to date. However, most approaches today lack a comprehensive review to examine DT benefits by considering both engineering product lifecycle management and business innovation as a whole. To fill this gap, this work conducts a state-of-the art survey of DT by selecting 123 representative items together with 22 supplementary works to address those two perspectives, while considering technical aspects as a fundamental. The systematic review further identifies eight future perspectives for DT, including modular DT, modeling consistency and accuracy, incorporation of Big Data analytics in DT models, DT simulation improvements, VR integration into DT, expansion of DT domains, efficient mapping of cyber-physical data and cloud/edge computing integration. This work sets out to be a guide to the status of DT development and application in today’s academic and industrial environment.


Introduction
With industries advancing into the Industry 4.0 era, factories are shifting towards a smart manufacturing paradigm with multi-scale dynamic modeling, simulation and intelligent decision making to enhance production capabilities (Davis et al. 2012).DT technology is an effective tool to fulfill the requirements of smart manufacturing by reflecting the physical status of systems in a virtual space (Tao et al. 2018c;Zheng et al. 2018d).Under the broad spectrum of CPS, the DT paradigm aligns well with a lifecycle-centered perspective (Schneider et al. 2019).DT technology is increasingly prominent as the focal point of the enhancement and evolution of global manufacturing.
As DT technology becomes more sophisticated, Liu et al. (2019b) described it as one of the strategic directions for manufacturing enterprises to progress.Designed to improve manufacturing efficiency, DT is a digital duplications of entities with real-time two-way communication enabled between the cyber and physical spaces (I-Scoop 2017).By providing a means to monitor, optimize and forecast processes, DT is envisioned by El Saddik (2018) as an approach for continuous improvement towards human well-being and quality of life.The maturity of this technology has also attracted attention from a wide range of industries, including healthcare and urban planning.City planners, assisted by DT technology, are able to interact with a data-rich city simulation, laying the foundation for a smart city as seen in the case of Singapore (Dassault Systèmes 2018).Gartner, a prominent global research and advisory firm describes DT as one of the top ten strategic technology trends in 2019 (Gartner 2019).Meanwhile, Grand View Research forecasts the DT market to grow to USD $27.06 billion by 2025, an approximate tenfold increase from USD $2.26 billion back in 2017 (Research 2018).With DT technology's ability to provide new possibilities for the emergence of new services and BMs, Industry 4.0 is no longer a "future trend" and many leading organizations have made it the center of their strategic agenda.For instance, with DT simulations and optimized decision-making, new insights can be obtained to produce smart products with self-awareness (Posada et al. 2015).Enterprises that are able to capitalize on this will benefit from the competitive advantages that are available to early adopters (Ghobakhloo 2018).Mabkhot et al. (2018) described an enormous range of benefits ranging from product design and verification, product lifecycle monitoring to shop-floor design, optimizing manufacturing processes and maintenance.Xu (2017) pointed out the role of DT technology in making smart machine tools via optimal decision support and machine health awareness analysis.The versatility of DT technology allows it to form the bedrock of future technologies, for example, it has the potential to be provisioned as a cloud service in support of cloud manufacturing.
In this review paper, past and present contributions to DT are analyzed by systematically examining the state-ofthe-art research articles from their engineering PLM and business innovation perspectives.The different industries and stakeholders involved with DT technology as well as tools and models utilized are investigated to provide a clear understanding on the various trends and directions this technology is heading towards.The rest of the paper is organized as follows: Sect. 2 outlines a literature review of DT concepts and related works including the systematic search process for relevant journal articles.Section 3 highlights the key technological tools and models used in DT creation.Section 4 describes the role of DT technology along stages of engineering PLM, while Sect. 5 discusses the business advantages of DT.Section 6 explores future perspectives of DT technology advancement and lastly, Sect.7 summarizes the contributions of the work done.

Literature review
This systematic literature review specifically focuses on works related to business and engineering aspects of DT technology.According to (Cook et al. 1997), a systematic review differs from traditional general review in that a duplication of the distinct and objective process is possible.As DT technologies are progressively developed for a wider range of industries to tackle extensive corporate functions such as strategic planning, it is essential that the technical, engineering PLM and business aspects of DT technology be reviewed to investigate the collective insights on theoretical analysis of existing studies.

Methodology in research selection
A systematic literature search was conducted in the Scopus database, covering most of the peer-reviewed interdisciplinary research papers, where a broad sum of studies on DT and other related literature can be identified using the systematic review methodology.Articles collected were further refined through a three-step approach (Reim et al. 2015), as depicted in Fig. 1.
Step 1 Publications identification and screening.
The first step serves to obtain quality publications via practical screening criteria during the past 5 years.Conference articles, working papers and commentaries are excluded to derive quality publications (Seuring and Müller 2008).Meanwhile, several keywords closely related to DT were identified.In addition to "Digital Twin", search terms such as "cyber twin" and "virtual twin" were used, as Oracle (2017) indicated that DT is made up of Virtual, Predictive and Projective twins.Although the differentiation represented different technological levels of DT, the purpose of these papers fit the scope of the study.The search phrase can be duplicated with the following searching sentence: "Topic = (Digital Twin OR "Virtual Twin" OR "Cyber Twin); Time Span: 2015-2019; Language: English; Type = "Article" (searched on 15/09/2019).This inclusive search yielded 256 relevant articles for further analysis.
Step 2 Theoretical screening process.
To emphasize both engineering PLM and business perspectives, articles advancing and applying DT technology are included, regardless of present or future considerations.More specifically, the selection benchmark is shown below: • DT applications and scenarios are selected, including using DT for situation optimizations.These studies involve conceptual and empirical discussion on DT implementations, allowing key technical aspects to be highlighted.• DT reviews and frameworks were examined to provide a comprehensive overview of trends, business functions and technologies involved.The insights gained from these studies will aid in identifying challenges faced for the evolution of DT. • Studies directly and indirectly involving DT concepts and challenges were examined, even those without mentioning DT in the title, keywords or abstract.This allows identification of future DT perspectives for new industrial developments.
Although various definitions exist, the core of DT remains the same.Therefore, in order to conduct a survey without any bias, DT is regarded consistently as a high fidelity virtual replica of the physical asset with real-time twoway communication for simulation purposes and decisionaiding features for product service enhancement, as depicted in Fig. 2 concurring with the contrast between DT and CPS as analyzed by Tao et al. (2018aTao et al. ( , 2019)).
Step 3 Reference analysis.
In this last stage, cited references from the original 110 articles that met the selection benchmark were further leveraged as a secondary source for literature analysis, resulting in an identification of 13 additional articles.Hence, this systematic literature review consists of 123 articles in total.For article analysis, DT categories were created based on their association with the research focus allowing easy reference and the categories were collated to form discussion themes.Additionally, 22 supplementary references were added to the reference section to make the survey concrete.

Evolution of DT
DT was first introduced by Grieves (2014) during his presentation of PLM in 2003.Although the initial concept was vague, a preliminary form of DT included both physical and virtual products and their interconnections.First serving as an inexpensive means to simulate varying conditions for NASA rockets, DT has since advanced technologically and expanded its scope of utilization.From the literature to date, DT-related enabling techniques have experienced exponential growth over time and its core idea has been transformed into distinctive concepts outlined in Table 1.

Descriptive analysis
As sensors become cheaper to procure and communication technology advances, DT provides a means to simulate and investigate scenarios that are otherwise too costly to explore. Figure 3 illustrates an exponential increase in DT utilization over the last 5 years with an expected increase in potential DT applications.
Table 2 shows the top journal names published in this area.IEEE Access is the most dominant source, accounting for 10% of articles reviewed, followed by other journals, especially in the manufacturing or industrial engineering  Stark et al. (2017) "Digital representation of a unique asset that compromises its properties, condition and behavior by means of models, information and data" Söderberg et al. (2017) "Using a digital copy of the physical system to perform real-time optimization" El Saddik ( 2018) "Digital replications of living as well as non-living entities that enable data to be seamlessly transmitted between the physical and virtual worlds" Zhuang et al. (2018) "Virtual, dynamic model in the virtual world that is fully consistent with its corresponding physical entity in the real world and can simulate its physical counterpart's characteristics, behavior, life and performance in a timely fashion" Qi and Tao (2018) "Virtual models of physical objects are created in a digital way to simulate their behaviors in real-world environments" Xu et al. (2019) "Simulates, records and improves the production process from design to retirement, including the content of virtual space, physical space and the interaction between them" Kannan and Arunachalam ( 2019) "Digital representation of the physical asset which can communicate, coordinate and cooperate the manufacturing process for an improved productivity and efficiency through knowledge sharing"   (Cheng et al. 2018).DT categorizes under Industry 4.0 and Fig. 4 shows the number of DT related research to highlight the enthusiasm of nations embarking on the DT trend.

DT techniques
As new industries acquire DT in a bid to boost productivity, efficiency and competitiveness, a diverse mix of tools and methodologies are used.This section provides a comprehensive analysis on tools and models used to create DT. Figure 5 shows the technology stack for DT establishment.Starting with data management and connectivity, models for DT communication are discussed.Subsequently, data representation and storage tools, machine learning tools and analytical methodologies are summarized.Lastly, microservices used to fulfill specific DT tasks such as virtual reality shop-floor are examined.Microservices are vulnerable to cyber-attacks, which could jeopardize the safety and quality of manufacturing systems.To raise awareness to these threats, (Elhabashy et al. 2019) analyzed cyber security issues in CPS, identifying attack methods and their impacts to operations.

Communication
Data acquisition and transmission are crucial in DT for realtime information flow and connectivity.This section emphasizes on key network architectures, data exchange protocols, as well as middleware platforms used in studies to facilitate information exchange and streaming processing.Network architecture involves integration of protocols and layered network interface through function blocks.Table 3 highlights the prominent architectures discussed such as multitier architecture and others.The OSI model, consisting of 7 layers (physical, data-link, network, transport, session, presentation, application), then establishes the concept of layered network architecture with the use of abstraction layers.These communication protocols are crucial rule sets for machine-to-machine connectivity between communicating entities.Table 4 highlights data exchange protocols in manufacturing environments used by data acquisition systems for high level DT communication.For ease of reference, the protocols are classified according to their nearest OSI model layers after which middleware platforms manage diverse software components for further development and streaming processing.Freeman (2016) described data stream processing system as analytics and continuous queries on real-time data.Table 5 summarizes key middleware platforms to enable seamless connectivity without altering infrastructures, allowing easier DT adoption into the current manufacturing ecosystem.These data acquisition systems are crucial for DT implementation in production environments with data collected via volatile (equipment specification, bill of materials etc.) and non-volatile data capturing processes (real-time sensor-based processing systems) (Uhlemann et al. 2017).

Representation
Heterogeneous data and domain knowledge gathered from shop-floor processes need to be modeled and integrated into manufacturing systems.Highlighted in Table 6, knowledge representation tools for DT creation such as ontologies and NoSQL databases are potential choices for achieving knowledge-based systems.Ontologies are favored as they address integration and domain-specific modeling concerns as well as reusing and sharing of knowledge.Knowledge representation languages such as OWL and knowledge management models such as RDF form the bedrock for DT creation while semantic integration of sensor data is explored to create taxonomies, ontologies and standards.Table 7 shows prominent data formats used in the research articles.

Computation
After selecting a storage engine, computational models are employed for batch-oriented and real-time data processing.Extracting practical knowledge from heterogeneous data is challenging and thus, determining the right methodologies and tools for querying and aggregating sensor data is crucial to DT construction.Machine learning and data processing tools provide a wide range of solutions ranging from analytics to automation and these provide DT with decision-aiding capabilities via enabling tools such as computer vision.Table 8 summarizes the computational processing tools used in this review.In Table 9, machine learning and analytics methodologies, including statistical kits for optimization are presented.Due to the overlapping nature of DT applications

Microservices
Microservices are software development tools constructed as a set of loosely coupled services.Thönes (2015) describes this architectural style as an enabling feature for an application to be built as a suite of relative small, consistent, isolated and autonomous services performing specific tasks.
Based on RAMI 4.0 (Rojko 2017), Table 10 provides a list of virtualization tools in modern production systems, to allow monitoring and tracing services of shop-floor assets for automated conflict resolution and performance enhancement through decision-aiding support and control.Table 11 highlights tools used in model creation and DT simulation of high fidelity asset replicas while Table 12 highlights validation tools provide support for task verification to ensure data accuracy and integrity.

DT perspectives on engineering PLM
DT perspectives in PLM stages are analyzed with the adoption of a generic competitive process framework proposed by Casadesus-Masanell and Ricart (2010) that outlined DT's implementation structure and process.By adopting this framework, companies can focus on relevant PLM aspects for product enhancement.The review highlights relevant industry applications and provides an overview of DT capabilities.
In engineering, Nasir et al. (2016) described PLM as a process of managing the product lifecycle from inception till disposal.PLM integrates people, data, processes and systems to provide product information support.Figure 6 provides a breakdown on the 54 DT papers identified to involve engineering PLM phases.Generally, an engineering PLM has 5 sequential stages (Stark 2016) and Table 13 outlines the various stages involved.

Design stage
In engineering PLM, DT frameworks and technologies enhance the design stage in a responsive, dynamic and comprehensive manner.This section analyzes DT capabilities for product improvement.
DT-based design and production integration DT approaches were used to integrate product design with production.Guo et al. (2018b) used a modular approach to  2019) adopted the open source approach for the design of a DT demonstrator and while Alam and El Saddik (2017) identified basic and hybrid computation-interaction modes with a DT architecture reference model in a telematics-based prototype driving assistance application.
Service innovation Service innovation is demonstrated by Zheng et al. (2018a) with a personalized smart wearable design via Smart PSS and DT to achieve user satisfaction with minimal environmental impact.Driven by smart connected devices, users can take part in the co-development of future products via cloud computing (Zheng et al. 2018b).
Analysis and validation through DT To deal with geometric reconstruction problems, Biancolini and Cella (2018) presented a mesh morphing workflow based on radial basis functions for model validation via DT.

Manufacturing stage
DT wields large influence in the Manufacturing stage with a wide range of novel and innovative studies aiming to make production process efficient, reliable and adaptable.
Production digitalization To react better to shifting consumer trends, DT is used to digitalize process models.Lu and Xu (2019) introduced a cloud-based manufacturing system architecture to achieve on-demand production, thus achieving better business flexibility.Modeling techniques were studied for DT construction as Liu et al. (2018a) created a machine tool cyber twin, achieving better connectivity and flexibility.Bao et al. (2018) proposed a DT modeling and operating construction approach in an aircraft structural parts machining cell case study while Tan et al. (2019) proposed a DT construction framework which models IoT data into a simulation.For DT application in shop-floors, Ding et al. (2019) used DT technologies to enhance interconnection and interoperability between cyber and physical shop-floors whereas Zhang et al. (2018) presented a novel production system architecture that also supported job scheduling in an aircraft engine manufacturing case study.Modeling strategies To enhance output, DT modeling methodologies are built to suit diverse conditions.Luo et al. (2018) proposed a multi-domain unified modeling method as a cyber-physical mapping strategy also used for fault prediction and diagnosis.Zheng et al. (2018c) introduced parametric virtual modeling and construction flow of DT application subsystems to fulfilled the case of a welding production line.In an aircraft assembly context, (Guo et al. 2018a) improved competitiveness with digital coordination model, utilizing DT to accomplish better flexible assembly accuracy and efficiency.Sharif Ullah (2019) created a semantic modeling methodology to compute virtual abstractions for material removal processes.
Production optimization Optimization of production aspects such as manufacturing speed and machine control were studied.In the dyeing and finishing industry, Park et al. ( 2019) proposed a service-oriented platform to enhance performance measures and achieve cost reduction through optimization algorithms.Moreno et al. (2017) showcased a DT for a sheet metal punching machine to optimize NC machining programs while, Zhao et al. (2019)  DT-enabled situational adjustments DT allows adjustments towards practical situations and simulations to deal with production process irregularities.Sierla et al. (2018) introduced a DT concept for automated assembly planning and asset coordination in a manufacturing cell.Lu and Xu (2018) encouraged DT adoption with a test-driven resource virtualization framework to virtualize complicated factory setups.
Monitoring production process Operation monitoring and virtualization require vast amount of data.Angrish et al. (2017) described an architecture based on NoSQL to store and handle streaming data in a scalable and flexible database in order to control virtualized production assets.Morgan and O'Donnell (2018) demonstrated a cyber-physical monitoring process of a CNC machine using a range of real-time sensor input that serves as a platform for future DT development.

Distribution stage
In logistics, DT uses real-time tracking and other solutions to facilitate operations.As industry players shift towards smart warehouses with industrial robots, DT is utilized to enhance warehouse safety and efficiency.
Robot-human collaboration Industrial robots are highrisk entities from a safety standpoint and DT assists in understanding and managing these robots to reduce health risks and reassure employees.For instance, Petković et al. (2019) proposed a Theory of Mind-based algorithm to perceive human reactions to robot assistants operating in changing environments via virtual reality DT.Nikolakis et al. (2019) implemented a DT approach to enhance planning and control using simulations to analyze productivity in logistics operations.Bilberg and Malik (2019) presented a DT-driven assembly system to demonstrate robotic automation with human flexibility.
Warehouse management DT can optimize warehouse management systems by providing decision-aiding support and comprehensive outcome analytics.Bottani et al. (2017) constructed a DT for job-shop production system involving scheduling for automated guided vehicles to transit in a logistics environment.Baruffaldi et al. (2019) illustrated a novel warehouse management decision-support tool by addressing factors such as clients' data, cost and returns on investment uncertainty.
Supply chain optimization Defraeye et al. ( 2019) slashed perishable losses with the aid of DT by improving the refrigeration process and logistics during distribution.

Usage stage
DT's capability in the usage stage involves predicting and designing next generation products, product upgrading and supporting the upkeep of manufacturing assets.By utilizing data and analytics from sensors embedded in smart products and tools, operations, reconfigurations and maintenance processes can be improved.
Knowledge reuse and evaluation DT provides decisionmaking support for multi-dimensional processes, strategy improvisation and process forecasting via knowledge recycling and awareness.Liu et al. (2018b) proposed a DT process reusability evaluation approach to prototype diesel engine models.Arafsha et al. (2019) introduced a modular framework for DT creation through action monitoring and data analytics.
Workflow improvement DT enhances conventional engineering analytics with information integration for a digitalized product life cycle.Iglesias et al. (2017) aimed to enhance engineering analysis workflows to enhance JET divertor operations with the DT approach.Haag and Anderl (2018) demonstrated a concept in which a DT will be developed alongside the product and remain its virtual counterpart throughout the entire product life cycle.Schneider et al. (2019) presented a virtual engineering method, integrating DT paradigm with lifecycle approach in a desalination plant case study.
Shop-floor enhancement Shop-floors are commonly associated as hives of activity in which DT can serve to improve the assembly layout, manage asset flow and integrate data to enhance production.Tao and Zhang (2017) constructed a shop-floor DT and discussed key components towards a trend of new paradigm directed towards smart and connected shop-floors.Zhuang et al. (2018) proposed a smart production management and control approach of product assembly shop-floors with a satellite assembly case study.
Digitalization of plant management Digitizing plant infrastructure provides a comprehensive overview of the various inefficiencies plaguing the system whereby DT analytics and solutions can ensure operational reliability.GE (2016) used DT to monitor and optimize power plant performance and showcased the capability to balance and optimize trade-offs between uncertain factors.He et al. (2018) demonstrated a cross-technology communication application to provide monitoring and decision-making support for an ultra-high voltage converter station case study.
Increasing energy and resource efficiency Reducing consumption is a key concern and DT is able to develop smart analytics models to enhance operational efficiency.Kannan and Arunachalam (2019) developed a predictive model for redress life identification and computation with a DT grinding wheel case study.To improve fuel efficiency, Coraddu et al. (2019) proposed a DT method to measure the influence of fouling on ships.MacDonald et al. (2017) leveraged simulation from sensor data to predict failures and diagnose inefficiencies in an operating pump demonstration while Ferguson et al. (2017) used a Siemens PLM software to simulate digital performance of water pumps, employing DT technologies to improve existing and next-generation products.
DT-driven PHM As real-time monitoring and simulations pave way for predictive maintenance, Xu et al. (2019) presented a DT fault diagnosis method using deep transfer learning in development and maintenance for a car bodyside production case study.Tao et al. (2018b) demonstrated a PHM method with an equipment DT, making use of system interaction and data fusion in a wind turbine case study.Wang et al. (2015) combined high-performance fatigue mechanics with filtering theories for aircraft diagnostics and prognostics while Tao et al. (2018a) conducted a review, focusing on DT-driven PHM techniques and applications as an enabling technology for smart manufacturing.Xia and Xi (2019) explored PHM methodologies for cyber-physical systems involving monitoring, data representation and computations, setting the stage for future DT applications.

End-of-life stage
Termed reverse logistics by Govindan and Soleimani (2017), this stage aims to reduce harmful repercussions on human and environment by emphasizing on disposal, remaining lifetime prediction, smart recycling and material recovery.Lu et al. (2019) proposed a DT approach for engine remanufacturing suited for small scale operations while Wang and Wang (2019) developed DT product models to facilitate the recycling of electronic equipment.Using DT, Popa et al. (2018) presented a novel approach to design a glass panel recycling flow and establish a process installation architecture which achieved a higher glass recovery rate.
Figure 7 shows a summary of DT benefits for each life cycle stage.With DT technology bolstering life cycles of products (Lee et al. 2016), control systems and resources can be put in place to intervene at the right moment on the right assets.The next section shows how DT is able to influence business aspects to increase profitability.

Business innovation perspectives
A growing number of industries are looking to improve profitability from cyber-physical technologies.Baden-Fuller and Morgan (2010) defined BMs as the value developed and delivered to clients.Adrodegari et al. (2017) explained about value monetization using BMs, describing it as a management method that bolsters critical decision-making.This section highlights the benefits received from DT adoption.Figure 8 adopts a combined set of BMs proposed by Wirtz et al. (2016).In reality, such rigid configuration is not always achievable and therefore, only considered as interrelated.
BMs affect different stakeholders when employed, which in this review, refer to the main beneficiaries upon successful implementation of DT.Identified stakeholder categories are shown in Table 14, representing DT stakeholders in manufacturing ecosystems.In this review, no articles were found to involve the network model and procurement model, since DT technologies do not aid external interactions to influence joint value creation and achieve cost-effective procurements.Although there is a growing trend of DT usage in construction, healthcare and other unconnected fields, DT remain predominantly applied in the manufacturing industry currently.Thus, this section strives to shed light on the versatility of DT and highlight its potential to value add to manufacturers and enterprises via BMs.

Strategic components
Strategic components create value for the businesses via internal input factors and set the directions for optimal resource allocation in order to maximize profitability.The strategy model acts as a guide to influence development of BMs and comprises of policy making to capitalize on DT trends, thus maintaining the industries' relevance.Table 15 shows strategies undertaken by the various industries to incorporate DT into policymaking.Another part of strategic components is the resource model.DT optimizes resource allocation, enhance operational efficiency and increase product output.The same table shows DT's influence in the resource model with benefits including cost reduction, process monitoring and decision-making support for machine PHM.The industrial popularity of DT technologies proves that DT is versatile in many fields and the potential to value add to a large proportion of the stakeholder ecosystem.

Customer and market components
This component focuses on consumer experiences and convenience through better-suited products and satisfaction while exploring alternatives to increase competitiveness through DT.The customer model focuses on attaining customer satisfaction through better quality products and services, while enlarging client bases via new market access and co-creation initiatives as shown in Table 16.Another aspect of the component is the market offer model.Known as value proposition, the market-offering model's objective is to increase product value by taking into account competitors and the entire market structure.Lastly, with existing forms of revenue streams (markup, licensing, subscription etc.), DT's role in designing revenue stream and structure is presented.

Value creation components
Value creation components emphasizes on creating customer value through better quality products, cost effective procurements and detailed financial planning to attain a frictionless capital flow.In addition, value creation for stakeholders insures the future availability of investment capital for operations.The manufacturing model aims to improve product quality via internal company processes with Table 17 showing DT providing positive value creation to existing products and services.The financial model in the same table shows how DT supports budgetary management through cost structure analysis and detailed financial outlines.
By highlighting the application benefits of this technology from a management standpoint, this review offers guidance for future DT adopters to capitalize on the advantages and stand out from the competition.With the consumer market expecting highly personalized smart products to be offered as services, it is apparent that as Industry 4.0 revolutionizes the rules of business, conventional business and marketing strategies will become unproductive (Ghobakhloo 2018).Thus, in order to develop new strategies, the current levels of digital capabilities have to be evaluated so that enterprises can capitalize on the opportunities offered by DT.

Discussion
When Michael Grieves first introduced DT in 2003, it was a concept for product monitoring throughout the lifecycle.From the articles reviewed in 2017, developments on DT were directed towards establishing a real-time 2-way communication before evolving into a dynamic virtual entity with model simulations in 2018.Today, digital cooperation is emphasized with decision-aiding support to optimize production performance and maximize profitability.DT security and privacy concerns are envisaged to be a key discussion The technical aspect reveals DT tools used in smart manufacturing featuring overlapping DT methodologies and manufacturing procedures such as the NSGA-II algorithm and MES.These manufacturing perspectives reveal the various types of DT such as partial, clone and augmented DT to be created for different applications (Kucera et al. 2016), providing a road map for developers addressing specific issues.The engineering PLM aspect reflects a lack of focus on the end-of-life stage.Hence, further studies are required to transform the product lifecycle into a continuous cycle as part of smart manufacturing paradigm (Flumerfelt 2017).Remanufacturing DT hold a probable approach in reducing and reusing obsolete products due to environmental concerns.Technology-business integration displays different strategies and forecast for upper management to conduct sales effectively.With revenue models, quality products and co-creation, consumer satisfaction is achieved with paradigms such as mass individualization.In addition, product-service bundle offerings are gaining popularity, lowering principal costs and optimizing output by offering automated real-time situational recommendations.In combining and deploying relevant BMs, enterprises can enhance their unique industry forte with DT to stay competitive.For example, small businesses as highlighted by Lu and Xu (2019) and Park et al. (2019) were able to adopt new revenue and resource models with the aid of DT, proving that DT-enabled BMs are viable.While DT is not only low cost solution to increase business competitiveness, they also benefit stakeholders ranging from customers to management while allowing employees to adopt a supervisory role.
Vast expansion in application potential ensures the continuous evolution of the DT concept.Hence, it is important to understand and identify the areas of research that authors are embarking on.From the eight key DT future directions, improvement on DT quality such as mappings and simulations before embarking on novel industry applications is crucial as only with refined DT features can further utilization such as incorporating virtual reality, quality decision support be more effective.

Future directions
As DT technology advances, researchers have highlighted the future directions to encourage mass adoption by enterprises.To better comprehend researchers' views regarding DT future perspectives, the significance of each perspectives is described in the following.A modular approach allows the construction of flexible DT, resulting in new application modes while reducing development time.Realizing modeling consistency and accuracy will improve the quality of DT, enhancing the benefits of DT applications.Incorporation of Big Data analytics into DT will provide more insights, resulting in better decision-making support while improvements in DT simulations allows better monitoring and transparency during processes.Virtual Reality integration unlocks further advancements into relevant fields such as education while extending DT to other domains allow better assimilation with the company's strategic objectives and production process.Efficient mapping of cyber-physical entities enable effective mechanisms to support situational adjustments and reduce uncertainty.Lastly, cloud and edge computing integration allows DT to process at a faster pace while processing vast amounts of heterogeneous and semantic data.
The 8 future perspectives are summed up into the 3 essential DT perspectives to improve the comprehensiveness of this section.Technical aspect.Most authors believe that improving modeling and simulation accuracy in a standardized manner is a key direction towards a higher quality DT.Engineering PLM aspect.The PLM paradigm allows a broader application of DT as enterprises push towards green manufacturing with DT optimizing quality production throughout a full loop cycle.Business aspect.DT with Big Data capabilities allows management to make informed decisions via its decision-aiding functionalities.

Conclusion
In recent times, awareness on DT technology as an enabling tool for bridging physical and cyber world has been growing exponentially.DT has been exploited for a wide range of applications, resulting in various interpretations and developments without a unified concept.To bridge this gap, this paper presents a systematic survey of existing DT research published in the last 5 years.Findings on the key benefits of DT are outlined below and categorized into three broad areas.It is hoped that this will provide insights on its future applications for a wider range of stakeholders and industries.
Technical aspect DT creation and development requires extensive knowledge on different technologies to ensure seamless integration between heterogeneous components.To overcome this challenge, four essential categories of communication, representation, computation and microservices were identified, forming a technology stack to ensure a coherent and consistent DT implementation.Tools and models used by researchers to add value towards productivity and adaptability of DT systems are classified accordingly, to serve as a reference model for academics and industries in their exploration and applications of DT in the near future.
Engineering PLM aspect As DT technology is primarily applied in the manufacturing industry, the analysis of engineering PLM aspects aims to reflect the effectiveness of DT towards the handling of products as it moves through typical lifespan stages.The PLM stages are subdivided into specific advantages that DT brings to facilitate innovation and growth towards smart manufacturing.As green, social, individual, intelligent, service-oriented and other manufacturing characteristics have become the development requirements and trends of the future manufacturing industry, this engineering perspective brings forth a vision of sustainable product development by utilizing DT technologies to extend the cradle to grave process into a full loop cycle.By promoting awareness on the benefits of DT to enhance the effectiveness of production operations via quantitative methods, comprehensive analysis and application case studies such as robot-human collaboration, knowledge reuse etc., it becomes clear that realizing DT interactions between human, machine, objects and environment in simulation models and manufacturing processes will gradually become vital for production systems.
Business aspect DT brings forth a wide variety of benefits from a business perspective for both small and large enterprises.Three essential components, strategic, customer and market, and value creation were identified, encompassing BMs achieving value monetization when incorporated with DT.Industries and stakeholders are determined to provide a comprehensive analysis towards the benefits of DT, allowing upper management to envision a future, where DT plays an essential role in delivering value to consumers and maximizing profits.
Furthermore, this study analysed the trends and viewpoints of existing research and established eight objectives to improve current DT.In the near future, the standards for real-time two-way mapping between physical and virtual models are essential towards the development of an updated and transparent DT system, in order to achieve successful decision-making outcomes and increase users' trust.The authors hope that this research can be regarded as a guideline for more research and discussion on DT aspects towards smart manufacturing and Industry 4.0.

Fig. 4
Fig. 4 DT publication count categorized by countries

Fig. 5
Fig. 5 Technology stack for DT creation

Fig. 6
Fig. 6 Overview of engineering PLM phases addressed demonstrated a joint optimization DT model for coordinating micro punching processes to boost punching speed.Using geometric assurance DT,Tabar et al. (2019) reduced computation speed for weld points.Liu et al. (2019b) described a DT-based machining process evaluation method for a marine diesel engine manufacturing process.Liu et al. (2018d) researched on a DT hot rolling production scheduling model and provide decision-aiding support.Coronado et al. (2018) presented manufacturing execution system as a core DT technology for production control and optimization, allowing easy implementation and lowering costs.Söderberg et al. (2018) applied real-time geometrical quality control for welded components through DT to enhance production quality for a range of welding processes.Individualized production Besides enhancing manufacturing processes, DT allows the shift towards individualized production.Zhang et al. (2017) presented a DT-based approach to provide decision-aiding support and analytics for rapid individualized designing of a hollow glass production line.Söderberg et al. (2017) utilized DT in the shift towards individualized production by leveraging simulations to control and optimize manufacturing systems.Liu et al. (2018d) introduced the rapid individualized designing of automated flow-shop system through DT to provide design validation.Leng et al. (2018) presented a mass individualization paradigm using DT to conduct parallel controlling, providing proactive decision support.

Table 1
Definitions of DT "Virtual representation of what has been produced"

Table 2
Top journals presented

Table 4
Data exchange protocols for DT creation Wireless-HART Used in wireless sensor networking technology Transport Laaki et al. (2019), and Senthilnathan and Annapoorani (2018) TCP Main protocol for enabling two hosts to exchange data Liu et al. (2018a), and Ardanza et al. (

Table 5
Middleware platforms for software development

Table 6
Databases and data management for DT creation

Table 8
Computational processing for DT creation

Table 14
Definitions of the various stakeholders

Table 15
Table 18 identifies eight major aspects in which DT has room for further development.With increasing research done on combining DT with emerging technologies such as blockchain and virtual reality, applications in new fields such as infrastructure, education and healthcare are imminent.DT enables automation, accessibility and transparency while lowering principle costs such as resources and man-hours.While majority of the studies view application to other domains as potential future work, DT technical aspects are not well established enough to ensure success in other fields.

Table 17
DT's influence on value creation components