Conference PaperPDF Available

Digital Twins: A Meta-Review on Their Conceptualization, Application, and Reference Architecture



Content may be subject to copyright.
Digital Twins: A Meta-Review on Their Conceptualization,
Application, and Reference Architecture
Alexander Rossmann
Reutlingen University
Dieter Hertweck
Reutlingen University
The concept of digital twins (DTs) is receiving
increasing attention in research and management
practice. However, various facets around the concept
are blurry, including conceptualization, application
areas, and reference architectures for DTs. A review of
preliminary results regarding the emerging research
output on DTs is required to promote further research
and implementation in organizations. To do so, this
paper asks four research questions: (1) How is the
concept of DTs defined? (2) Which application areas
are relevant for the implementation of DTs? (3) How is
a reference architecture for DTs conceptualized? and
(4) Which directions are relevant for further research
on DTs? With regard to research methods, we conduct
a meta-review of 14 systematic literature reviews on
DTs. The results yield important insights for the current
state of conceptualization, application areas, reference
architecture, and future research directions on DTs.
1. Introduction
The concept of digital twins (DTs) is receiving
increasing attention in research and management
practice [1], [2]. Mentioned initially in 2003 by Michael
Grieves, DTs pave the way for cyber-physical
integration and serve as a bridge between the physical
world and the cyber world [3]. Grieves's initial
description defines a DT as a virtual representation of a
physical product [4]. Later, Grieves expands on this
definition by describing DTs as consisting of three
components: a physical product, a virtual representation
of that product, and bidirectional data connections that
feed data from the physical product to the virtual
representation and back [4]. Scholars have also provided
evidence for the notion that DTs not only deal with
physical products but also are applicable to any
Gartner Hype Cycle for Emerging Technologies 2019,
connection and bidirectional exchange between a virtual
and physical entity [5][7].
Initially conceptualized for manufacturing [3], [8],
the idea of DTs has reached multiple domains, including
smart cities [9], healthcare [10], management [11], and
psychology [12]. Since the inception of DTs in 2003, the
concept has grown in interest and was listed as a key
strategic technology trend for the first time by Gartner
in 2019.
A search of the term digital twin in academic
databases shows increased research interest in the topic.
Until 2017, the number of academic articles on DTs was
only in the single-digit range. Since then, the number of
yearly publications has grown exponentially [13].
The idea of DTs defined by cyber-physical
integration is creating a broad array of opportunities for
organizations. However, owing to the short and dynamic
development cycle, the concept of DTs in terms of a
clear conceptualization of the concept and its properties,
an overview of application areas and enabling
technologies, and reference architectures for the
implementation of DTs is still blurry [1], [2]. In
response to the ambiguity in current research, this paper
aims to answer four research questions: (1) How is the
concept of DTs defined? (2) Which application areas are
relevant for the implementation of DTs? (3) How is a
reference architecture for DTs conceptualized? and (4)
Which directions are relevant for further research on
The approach for the exploration of these research
questions follows established guidelines for a meta-
analysis on systematic literature reviews (SLRs) [14],
[15]. Therefore, we conducted a meta-review of 14
existing SLRs on DTs and analyzed the body of related
work with respect to the presented research scope.
Accordingly, are present and discuss the results of the
review. The results yield important insights for the
current state of conceptualization, application areas,
reference architecture, and further research on DTs.
Proceedings of the 55th Hawaii International Conference on System Sciences | 2022
Page 4518
(CC BY-NC-ND 4.0)
2. Related work
A search on Google Scholar and Web of Science
with the search strings (Systematic Literature Review)
AND (Digital Twin*) on the topic level leads to 14
articles, with related work published between 2018 and
2021. The following review is fueled by a full read of
these articles and a corresponding analysis of the areas
of application, time frames, databases, and article
quantity of each SLR, as well as an overview of central
concepts of each article.
Multiple reviews on the topic of DTs are present in
current research. Initially, the DT concept emerged in
the context of manufacturing and Industry 4.0 [16].
Kritzinger et al. [17] presented an SLR on DTs in
manufacturing in 2018. They noted that DTs are
commonly known as a key enabler for the digital
transformation, however, in literature is no common
understand concerning this term. In this review, the
authors adopt the definition of Tao et al. [18] that “the
digital twin is an integrated multi-physics, multi-scale,
probabilistic simulation of a complex product and uses
the best available physical models, sensor updates, etc.,
to mirror the life of its corresponding twin”. An
important contribution of Kritzinger et al. [17] is the
conceptualization of DTs with respect to the level of
integration between the physical and its digital
counterpart. The terms “digital model, “digital
shadow,” and “digital twin” are often used
synonymously. However, the given definitions differ in
the level of integration; that is, a digital model contains
a pure digital representation of an existing or planned
physical object that does not use any form of automated
data exchange, and a digital shadow comprises a one-
way data flow from the state of an existing physical
object to a digital object. Adopting this distinction,
Kritzinger et al. [17] categorized 53 articles published
between 2001 and 2017 with respect to the application
of digital models, digital shadows, and DTs.
Comparable reviews with a strong focus on
manufacturing and Industry 4.0 were conducted by
Cinar et al. [16] and Catarci et al. [19] with a rather low
number of articles (n = 19 and n = 16, respectively). A
major contribution of Cinar et al. [16] is their work on
GE digital with respect to a hierarchical classification of
DTs in four properties: component, asset, system, and
process. A component twin is a major sub-component
affecting the performance of the asset to which it
belongs. Asset twins can be a collection of component
twins, such as a motor or pump. System twins are a
collection of asset twins performing a system function
such as a production line. Process twins usually provide
a perspective to a set of operations at the highest level
and generally focus on processes rather than equipment.
The main contribution of Catarci et al. [19] is the
connection of concepts dealing with cyber-physical
systems (CPSs), the Internet of Things (IoT), and DTs
in smart manufacturing. They envision a system
architecture for DTs in which humans can specify a goal
and take advantage of DTs to automatically compose the
corresponding physical processes. An important aspect
of this view is the introduction of a service perspective
to DTs. Various properties of DTs are discussed (e.g.,
connectivity, autonomy, homogeneity, ease of
customization, traceability). Moreover, Catarci et al.
[19] define data models, patterns for data exchange, and
the ability to run simulations as core facets of DTs and
review current industry solutions for DTs in smart
manufacturing as Eclipse Ditto, Bosch IoT Things
solutions, GE Predix, Microsoft Azure IoT, Amazon
AWS IoT, and IBM Watson IoT.
A corresponding SLR in the area of smart
production is the concept of cyber-physical production
systems (CPPSs) [5]. CPPSs are systems of systems
with autonomous elements connected with each other,
on and across all levels of production, from processes
through machines up to production and logistics
networks, enhancing decision-making processes in real
time, response to unforeseen conditions, and evolution
along time. Wu et al. [5] reviewed 100 articles with
respect to a conceptual definition and engineering
development of the topic. They arrived at a concept map
with three conceptual categories of articles: need
analysis, concept exploration, and concept definition.
Furthermore, they defined five categories of articles on
an engineering level: smart connectivity, data-to-
information conversion, cyber, cognition, and
configuration. In particular, DTs in this framework are
part of the cyber level on the engineering side of CPPSs.
Lim et al. [20] provide a comprehensive overview
of the technology stack for DTs in smart manufacturing.
They also discuss potential contributions of DTs along
the whole product life-cycle management process and
expand perspectives to business models and business
innovation. Based on an SLR of 123 research articles
published between 2015 and 2019, the technology stack
of DTs covers multiple layers, including data
acquisition, data management and connectivity,
network architecture, data representation and storage
tools, data analytics and machine learning (ML),
microservices, and cyber security. Furthermore, Lim et
al. [20] discuss contributions of the DT concept on
different product life-cycle management stages as
product design, manufacturing, distribution, usage, and
end of life. This allows various fresh perspectives for
business model innovations, covering aspects of
corporate strategy, customer and market segmentation,
and value creation components.
Page 4519
With the further development of research on DTs,
increasingly more scholars are working on an extension
of the concept beyond manufacturing. An example of an
SLR in this direction is that of Zhang et al. [6]. They
reviewed 59 articles on DTs and services to develop a
framework for holistic industrial product service
systems. Within such systems, the application of DTs is
reviewed along the whole life cycle of products and
services from beginning to end of life. This includes a
notion on design, sales, distribution and usage, and reuse
and recycling of products and services.
A comparable review extends the analysis from
smart production to industrial services with an
additional focus on predictive maintenance and after-
sales services. Melesse at al. [21] summarize 25
research articles in the selected domains and argue that
DTs play an important role in operations throughout the
whole product life cycle, including the concept and
design stage, manufacturing planning and execution,
sales, product usage, maintenance, and product
Beyond the area of smart manufacturing, several
other SLRs extend the idea of DTs to a more general
level of conceptualization. Jones et al. [22]
characterized DTs through a review of 92 articles and
identified 12 characteristics. Core characteristics of DTs
include physical entities (e.g., vehicles, components,
products, artefacts); virtual entities (e.g., data, models);
the physical environment of a DT (e.g., factories, cities);
the virtual environment of a DT (e.g., databases, data
warehouse, cloud platforms, servers); parameters, as
types of data, information, and processes that are passed
between the physical and virtual entities; fidelity,
defined as the number of parameters, their accuracy,
level of abstraction, and transfer between the virtual and
physical environment; state, or the current condition of
both the physical and virtual twins or the current values
for each of the measured parameters; physical-to-digital
versus virtual-to-physical connectivity; twinning rate, or
the act of synchronization between the virtual and
physical states (e.g., real-time, near-time); physical
processes (e.g., smart factories, three-dimensional (3D)
printing, robot control, medical health applications); and
virtual processes (e.g., simulation, modeling,
A comparable SLR for general conceptualization
conducted by Enders and Hossbach [23] contains an
analysis of the dimensions for DT applications. With a
review of 152 research articles, they separated purpose,
industrial sector, the physical reference object for the
DTs, and other relevant dimensions for the description
of current applications. By 2019, most applications were
linked to manufacturing, while other sectors such as
automotive, aerospace, energy, healthcare, and smart
cities were also working with DTs. The main purpose of
most applications is simulation, directly followed by
monitoring and control. The idea of DTs is applicable to
various industries with multiple application areas.
Enders and Hosbach [23] also argued for a deeper
exploration of the DT concept in the information
systems domain and postulated the need for a
corresponding taxonomy. Van der Valk et al. [13]
describe such a taxonomy of DTs based on an SLR of
233 articles from different databases. A taxonomy
describes properties of a research object and relevant
differences due to specification of these properties in
research and practice. Relevant properties of DTs
include the data link between the physical and virtual
parts of the DTs (unidirectional, bidirectional), purpose
(processing, transfer, repository), the connection
between the physical and virtual parts (physically
independent, physically bound), accuracy of data
exchange (identical, partial), interface (machine-to-
machine, human-to-machine), synchronization, data
input (raw, processed), and the time the physical and
virtual parts are created (physical part first, virtual part
first, simultaneously).
In addition to the focused reviews in smart
manufacturing and holistic reviews on taxonomies,
properties, and general characteristics of DTs, other
SLRs have explored the concept within a specific
context or sub-topic. Dos Santos et al. [24] examined the
application of simulations for decision support in DTs.
In a review of 75 articles, they analyzed different
application areas and objectives for decision-making
through simulations with DTs. The main application
areas for such simulations are manufacturing, services,
logistics, healthcare, and constructions. Regarding
decision-making objectives, simulations with DTs are
used for production planning, process evaluation,
process control, and resource allocation. Dos Santos et
al. [24] also researched the applied platforms for
simulations such as Tecnomatix or Arena and relevant
software frameworks such as Python, Java, and
Stroboscope. Moreover, they evaluated different types
of connectivity between the simulation model and the
physical system (e.g., IoT devices, web services,
management systems), updating and synchronization
practices (real-time, near real-time), and the degree of
autonomy of the DT simulation model (e.g.,
autonomous command, recommended actions).
Another SLR with a more tapered orientation by
Rathore et al. [25] captures the role of big data and ML
in digital twinning. In a review of 61 sources in various
databases, patents, and technical reports, the authors
identified different applications of big data and ML in
the context of DTs in various industries. Examples of
potential ML algorithms and data models include
applications in production, healthcare, transportation,
education, and business. Furthermore, Rathore et al.
Page 4520
[25] proposed a model for the integration of IoT, big
data, ML, and DTs, in which (1) IoT and other data
sources create big data, (2) data are employed in data
models and ML algorithms, (3) simulations and
automation procedures are executed in the virtual
environment, and (4) such simulations and automated
processes are used for deployments in the physical part
of a DT. The SLR offers a detailed analysis of this model
with a description of the applied ML approaches in
different industries. Furthermore, the authors offer an
extensive overview on DT development tools,
evaluation procedures for the success of DT
applications, and reference architectures for digital
The concept of DTs is intensively integrated in the
development of CPSs, smart production, and the vision
of Industry 4.0. However, the basic idea behind DTs is
also applicable to other industries. Therefore, further
SLRs are available for DTs in smart cities [26] or the
enablers of and barriers to DTs in the process industry
[27]. In their gray literature analysis, Ketzler et al. [26]
compared the concept of DTs with established 3D city
simulations. They evaluated commonalities and
differences between the two concepts and argued that
DTs describe something more than a 3D city model
(including semantic data, real-time sensor data, physical
models, and simulations). Furthermore, they analyzed
current implementations of DTs in cities and concluded
that there are significant opportunities for up-scaling
DTs, with the potential to bring benefits to the city and
its citizens.
Finally, Perno et al. [27] conducted an SLR on
enablers of and barriers to the implementation of DTs in
the process industry. From a review of 38 articles, they
developed a framework for organizational capabilities,
several development issues, and performance effects for
DTs. As such, organizational preconditions such as
knowledge, design, and integration of a DT are
prevalent for implementation success. Moreover, the
development process itself contains several barriers
(e.g., lack of standardization, model development, data
quality, IT infrastructure). Finally, performance issues
need to be discussed at the very beginning of a DT
initiative (e.g., costs vs. benefits, reliability, robustness).
In summary, related SLRs on the concept of DTs
have initially appeared in the area of smart production
and Industry 4.0, with the distinction among digital
models, digital shadows, and DTs [17]; a hierarchical
classification of levels for DTs in a production
environment [16]; a conceptual connection among
CPSs, IoT, and DTs; and emerging perspectives on
system architectures and services based on DTs [19].
Within larger systems of systems, DTs are the digital
part of CPPSs [5]. The technology stack and several
implications of DTs along the whole product life-cycle
management chain, business models, and business
innovation are discussed by Lim et al. [20]. Several
scholars have expanded the idea of DTs beyond
production. This covers a discussion on the role of DTs
in industrial product service systems [6], [21]. Beyond
smart production, SLRs tend to develop general
characteristics of DTs applicable in multiple domains
[22]. Therefore, the concept of DTs is applicable to
various industries and application areas such as
automotive, aerospace, energy, healthcare, and smart
cities [23]. A taxonomy can be used to define general
properties and common differences in the specification
of DTs [13]. Beyond holistic frameworks for DTs are
SLRs with a strong focus on a specific area, such as the
application of simulations based on DTs [24] or the
connection among big data, ML, and DTs [25]. Rathore
et al. [25] offer an extensive review on the connection
between IoT and other data sources, data models, ML
algorithms, and data-based applications on the virtual
and physical part of DTs. Such frameworks and generic
reference architectures might be transferred to various
application areasfor example, to the implementation
of DTs for cities [26] or specific applications of DTs in
the process industry [27]. Table 1 summarizes the area
of application, time frame, database, quantity of
research articles, and the general concepts of the
different SLR on DTs.
Table 1. SLRs on the concept of DTs
Area of
Time frame,
ing, Industry
Scopus, n=53
ing, Industry
database not
defined, n=19
ing, Industry
4.0, digital
Scholar, n=16
ing, Industry
4.0, CPPS
Web of
ing, product
Page 4521
Area of
Time frame,
Scopus, n=59
Services, product
design, customer
purchase and
usage, reuse,
Scopus, Web
of Science,
Concept, design,
planning and
execution, sales,
product usage,
product renewals
General, DTs
Purpose, industrial
sector, physical
reference object
Direct, n=233
Taxonomy, data
link, purpose,
accuracy, interface,
data input, time of
making, CPS
Scopus, Web
of Science,
Scielo, IEEE,
Direct, n=75
Application areas,
(AI), ML, big
Hindawi, IGI,
Wiley, n=61
Standards, industry
integrated models
of IoT, big data, AI
and DTs,
evaluation of DTs,
development tools,
Scopus, Web
of Science,
Smart cities,
3D city
3D city models,
application of DTs
in cites, challenges,
3. Results
We present the results of the meta-review of SLRs
in the area of DTs according to the formulated research
questions. Therefore, we organize the following sub-
sections around conceptualization, application areas,
reference architecture, and future research directions for
3.1. Conceptualization of DTs
The conceptualization of DTs is an important
subject of multiple reviews and coined by the area of
application. Early articles on the subject focused on
smart manufacturing, physical production, and products
[16]. Therefore, such conceptualizations import
domain-specific aspects into the definition of a DT.
However, given the increasing application of the
concept in multiple domains, a general
conceptualization without any domain-specific
properties is required. Therefore, a mutual
understanding of a broad array of reviews defines a DT
as a CPS with physical and virtual (digital) parts. As
Grieves [4] argued, DTs serve as a bridge between the
physical world and the cyber world. Data flows between
a physical and a digital object with full integration in
both directions can be viewed as a central property of
DTs [17]. This feature leads to multiple implications, as
twinning demands not just a simple image of the
physical object but also a real interaction between the
physical and digital parts of the twin. Thus, the
conceptualization of DTs needs to be extended by
various other properties such as data models,
connectivity, accuracy, and synchronization [13], [24].
Moreover, the development of a DT is driven by a
specific purpose and expected benefits [6], [23]. DTs are
only valuable if the processing of real-world data in the
digital part leads to relevant insights and corresponding
services in the physical part of the twin. This leads to the
assumption that certain applications in the area of AI
(e.g., ML algorithms) should be treated as constitutional
property of a DT [25].
Given the mentioned properties, DTs need to be
conceptualized as a CPS, with an interactive
relationship between the physical and digital parts,
purpose, data connectivity in both directions,
corresponding data models, task-specific levels of
model accuracy and data synchronization procedures,
embedded AI in the digital part of the twin, and
dedicated services in the physical part.
3.2. Areas of application
Such a general conceptualization of DTs is fruitful
for multiple application areas. Originally
conceptualized for manufacturing [3], [8], over time, the
idea of DTs has reached multiple domains, including
smart cities [9], healthcare [10], management [11], and
psychology [12]. Therefore, the general idea of DTs is
not exclusively linked to a specific domain; rather,
digital twinning provides a framework for applications
in multiple domains and industries with a focus on two-
way interactions between a physical and digital entity.
Such a generalization of the concept is already
Page 4522
embedded in the analyzed SLRs, in which several
authors developed approaches for DTs in smart
manufacturing [16], [17], [19], but also in CPPSs with a
focus on services in an extended production
environment [5], stakeholder-specific services
throughout the whole product life cycle [6], or ideas for
the application of DTs in marketing and sales [21].
Thus, this meta-review provides evidence for a
systematic expansion of the concept in further domains.
It is likely that this expansion will continue over the next
years with diverse and fruitful applications for DTs in
multiple domains.
3.3. Reference architecture
Relevant sub-concepts for a general reference
architecture for DTs are present in multiple SLRs within
this meta-review [13], [22], [25], [28]. Rathore et al.
[25] provide a holistic approach for a general reference
architecture for DTs with multiple layers. The different
layers constitute a hierarchical order and include
multiple forward and backward interactions.
Furthermore, a distinction between different layers
supports the definition of properties and relevant
technologies per layer. Essential components for such
an architecture are also presented in Lim et al.’s [20]
SLR. Table 2 gives an overview of a nine-layer
architecture and corresponding properties. Space
restrictions prevent the inclusion of corresponding
technologies and application examples for each layer of
the architecture in Table 2. However, more precise
specifications are available in the assigned SLRs per
The bottom layer for the architecture of a digital
twin comprises physical entities [22], which include
vehicles, components, products, machines, streets,
parking lots, buildings, and other physical objects. The
commonality in these entities lies in their real-world
existence. To encompass all types, and in line with
existing literature [22], we propose the use of the term
physical entity for general applicability, where a
physical entity exists regardless of whether it has been
twinned, and the more specific term physical twin for
when the physical entity is twinned.
The second layer of the architecture includes
different strategies for data generation [20]. This
requires an identification of the physical entities for
digital twinning and the relevant parameters to be
generated. Parameters refer to the types of data and the
information that can be generated with the data. Typical
technologies for data generation include sensors or log-
files. In addition, fidelity is a relevant property defined
on the data generation layer. The term fidelity
describes the number of parameters, their accuracy, and
the level of abstraction that is transferred between the
virtual and physical twin [22]. The definition of fidelity
describes the required accuracy for digital twinning,
such as whether a minor part of the physical entity is
twinned [29] or the DT is a full mirroring of the physical
characteristics and functionalities [30].
The third layer of the reference architecture
includes network and connectivity. This layer covers
data acquisition and transmission as crucial elements for
real-time information flow and connectivity [28]. The
layer emphasizes network architectures, data exchange
protocols, and middleware platforms to facilitate
information exchange and streaming processing.
Network architecture involves integration of protocols.
Such communication protocols are crucial rule sets for
machine-to-machine connectivity between
communicating entities. In addition to network
architecture, this layer covers the connectivity
infrastructure applied in the DTs (e.g., Wi-Fi, Bluetooth,
ZigBee, mobile radio communication).
The fourth layer comprises data storage, data
integration, and big-data processing. Heterogeneous
data sources and domain knowledge gathered from
application processes need to be integrated in
operational database systems and an integrated data lake
On the fifth layer, the integrated data need to be
interpreted and prepared. Knowledge representation
tools for DT creation such as ontologies are potential
choices for achieving knowledge-based systems.
Ontologies are favored because they address integration
and domain-specific modeling concerns as well as the
reuse and sharing of knowledge [20]. Knowledge
representation languages such as the W3C web ontology
language (OWL) and knowledge management models
such as the resource description framework give the
foundation for DT creation, while semantic integration
of sensor data is explored to create taxonomies,
ontologies, data formats and standards.
Accordingly, the sixth layer is crucial for DTs and
deals with data models, algorithms, virtual entities, and
virtual twins. ML and data-processing tools provide
multiple solutions ranging from analytics to automation,
and these provide DTs with decision-aiding capabilities
via enabling tools [20].
The seventh layer is dedicated to micro-services
and the deployment of data models in real-life processes
[20]. Microservices are software development tools
constructed as a set of loosely coupled services. This
architectural style can be described as an enabling
feature for an application to be built as a suite of
relatively small, consistent, isolated, and autonomous
services performing specific tasks [31]. Microservice
architectures are available for different domains; for
example, in the application area of smart production,
RAMI 4.0 [32] provides an overview of layers and
Page 4523
microservices in production systems to allow
monitoring and tracing services of shop-floor assets for
automated conflict resolution and performance
enhancement through decision-aiding support and
The eighth layer deals with system security and data
privacy. As DTs affect real-life processes, security is a
major and cross-architectural issue within the design of
twinning concepts. Owing to the integration of sensitive
data in the case of stakeholder-related services (e.g.,
after-sales, product and content customization), privacy
is also becoming a relevant topic. Privacy-preserving
approaches in DTs can be classified into two categories:
cryptographic approaches featuring encryption schemes
and cryptographic primitives and noncryptographic
approaches with a policy-based authorization
infrastructure [27]. Properties of a reference architecture
should be able to detect security and privacy concerns
and minimize breaches and associated risks to which
DTs can be exposed.
Finally, the ninth layer pertains to the generation of
business models. This layer is of great importance
because DTs only make sense if they drive business
innovations and lead to compelling customer experience
and business models [20]. Although interest is growing
in multiple application areas, to date, DTs remain
predominantly applied in the manufacturing industry.
However, even in smart manufacturing, only a few
scholars have focused on the business model
implications of DTs [33], [34]. DTs strive to enhance
customer experience through better-suited products and
services. Attaining customer satisfaction through better
quality products and services, while enlarging the
customer base via new market access and co-creation
initiatives, drives the business model for DTs in smart
manufacturing. DTs also drive the development of new
products, services, and value propositions.
In summary, the twinning process begins with the
collection of data from the physical entities or with the
usage of already-collected data in the virtual twin (using
computer-aided software and/or simulations). The data
are subject of analysis and decision-making, in which
statistical and probabilistic approaches or mathematical
models are employed to develop the DTs. Throughout
the entire process, various big-data processing tools that
allow parallel processing on multiple nodes may be
employed. The overall data flow for the development of
a purpose-driven DTs begins with the creation of a data
model. After the data model is developed and tested,
data from both the physical and virtual twins are used to
deploy specific services to achieve the given
organizational objectives, such as design optimization,
dynamic process planning, or content customization
The presented architecture shows common ground
with existing frameworks for specific domains; for
example, the Open Systems Interconnection model
consists of seven layers (physical, data link, network,
transport, session, presentation, and application) and
paves the way for layered network architectures with the
use of abstraction layers. Therefore, further research on
the reuse and unification of different reference
architectures on DTs is required. Table 2 summarizes
the layers, properties, and corresponding sources
presented in this research.
Table 2. DTs reference architecture layers
Sources, SLRs
9. Business
model, processes
Value propositions,
revenue streams,
cost structure,
[6], [7], [21],
[22], [23], [25],
8. System
security, data
Security, privacy,
risks, abuse,
7. Micro-services,
deployment, loose
coupling, interfaces,
[7], [21], [23],
6. Data models,
algorithms, virtual
entity, virtual twin
Data models,
algorithms, ML,
decision rules
[7], [13], [19],
[22], [24], [25],
5. Data
preparation, data
ontologies, data
labels, fidelity,
twinning accuracy
[5], [7], [13],
[17], [19], [22],
4. Data storage,
data integration
big-data processing
[5], [7], [13],
[17], [19], [22],
[24], [25]
3. Network,
architecture, data
exchange protocols,
platforms, streaming
[5], [7], [13],
[17], [19], [22],
[24], [25]
2. Data generation
Parameters, types of
data, fidelity,
[5], [6], [7], [13],
[17], [19], [22],
1. Physical
entities, physical
Physical existence
[6], [7], [17],
[19], [22], [25]
3.4. Future research directions
Another objective of this meta-review is to
summarize future research directions for DTs on the
basis of the findings of the incorporated SLRs. The
rapidly increasing DT popularity and scope, as well as
the involvement of IoT, big data, and AI technologies,
broadens the research challenges of digital twinning.
We categorize these challenges in the following eight
3.4.1. Concept development. Different application
areas develop DT concepts from their specific domain.
Page 4524
Therefore, specified interactions and interfaces between
various disciplines are required. This gives rise to the
issue of a multidisciplinary development of the DT
concept [5]. On the one hand, a general framework and
reference architecture is required to develop DTs on a
mutual background [13]. On the other hand, a broad
array of domain-specific DTs will promote more
detailed implementations in various application areas.
With increasing research combining DTs with emerging
technologies such as blockchain and virtual reality,
applications in new fields such as infrastructure,
education, and healthcare are imminent [7].
3.4.2. Business models. Research on the business
pre-conditions and performance effects of DTs is
relatively scarce. A wealth of research focuses on
technical needs. In the future, other perspectives should
also be considered, such as performance in terms of
customer experience and business models. Some
existing reviews and guidelines for requirements
engineering could be a potential starting point for
further research initiatives [5], [7].
3.4.3. Integration. With respect to the integration
of DTs, technologies, devices, data, processes, and
systems should be integrated together in a strategic and
operating environment. The reviewed SLRs have
addressed some integration issues, such as device
integration, system integration, and data integration.
However, full integration of DTs in organizational
processes and enterprise systems has not yet been
addressed [27]. Such systems can foster effective
decisions, improve business processes, and make the
enterprise more competitive. Therefore, the integration
of DTs in enterprise systems is one of the main issues
for successful implementation [5].
3.4.4. Data entry, data preparation, data
augmentation. Further research needs to clarify several
questions in the data-entry process, such as how the IoT
facilitates data harvesting from a physical twin (using
sensors), data integration, and data sharing with the
corresponding virtual twins. This process can incur
considerable costs. Sometimes, twinning can be more
costly than the asset itself, in which case it makes little
sense to create DTs. By contrast, the collected data are
vast, heterogeneous in nature, unstructured, and noisy.
Thus, further research on data pre-processing is required
to ensure its effective use [21]. Specifically, applying
data-cleaning techniques is necessary to organize and
restructure data entry [5]. Furthermore, controlling the
flow of such a large amount of data is a significant
challenge. Finally, to improve the accuracy of the DT
model, the underlying ML algorithms require a certain
amount of data for training purposes [25]. The data
acquisition problem is a significant challenge in the
realization of DT models in small and medium-sized
companies. Therefore, future research needs to consider
approaches for data incubators and data augmentation
3.4.5. Big data. The explosive growth of social
media and IoT technologies in the industrial sector has
led to the generation of a large amount of data. To this
end, big-data analytics requires advanced architectures,
frameworks, technologies, tools, and algorithms to
capture, store, share, process, and analyze the
underlying data. There is also a potential for cloud- and
edge-computing platforms to handle DT-related data.
Cloud- and edge-computing integration allows DTs to
process at a faster pace while processing vast amounts
of heterogeneous and semantic data [7], [25]. Further
research is required given the insufficient possibilities
for synchronization between the physical and digital
parts, the absence of high-fidelity models for simulation
and virtual testing in different scales, the difficulties in
predicting complex systems, and the challenges with
gathering and processing large datasets [21].
3.4.6. Data analysis, ML, simulations. Algorithms
for data analytics have played a major role in DTs for
decision-making. However, the selection of a particular
ML model with customized configuration is
challenging. Every ML approach has diverse accuracy
and efficiency levels with different applications and
datasets. Therefore, depending on the motive and
application of a DTs, the selection of the best ML
algorithm and features is challenging [25]. Realizing
modeling consistency and accuracy will improve the
quality of DTs, enhancing the benefits of their
applications [7]. Moreover, twinning processes might
not only start from the physical entity but also be based
on simulations at the level of the virtual entity.
Therefore, different questions on data simulations are
relevant for further research on DTs [24]. Another
research topic pertains to the implementation of ML
algorithms with respect to operations and continuous
deployment. Implementation of an accurate multi-scale
DT model of work-in-process scenarios is still
challenging because the real-time changes during the
twinning process are difficult to perceive and simulate
3.4.7. Standardization. Although many DTs have
been developed in various industries, the creation of a
complex and reliable DT demands standardization.
Currently, no single standard focuses solely on digital
twinning. The ISO/DIS 23247-1 standard has limited
information on digital twinning, and therefore DT
deployment challenges are continuing to grow as a
result of the lack of standardization [21].
Standardization efforts are underway by the joint
advisory group (JAG) of ISO and IEC on emerging
technologies [25]. Many specific architectures for DTs
are proposed, but integrated design standards need to be
investigated by designers who take all disciplines into
Page 4525
consideration simultaneously [5]. RAMI 4.0 provides
such a holistic view of important aspects in smart
manufacturing that different stakeholders need. It
combines three core dimensions in a cuboidal space
covering (1) the whole life cycle from development to
disposal and resource recovery, (2) multi-layer
integration from asset to business, and (3) the
connection from products to the IoT and services.
3.4.8. Security and privacy. Some concepts such
as human-, product-, energy-, city-, and defense-related
DTs, are considered critical and may require stringent
security and privacy regulations [21]. First, with the
involvement of IoT devices in digital twinning,
emphasis needs to be put on the security of the under-
lying communication protocols [27]. Second, the large
collection of asset-related data needs to be stored
securely, to prevent data breaches from insider and
outsider threats [25].
4. Discussion
This paper deals with a meta-review of 14 SLRs on
DTs published between 2018 and 2021. We analyze the
results of the meta-review with respect to
conceptualization, application areas, reference
architecture, and future research directions for DTs.
An evaluation of results leads to several important
implications for research and management practice on
DTs. First, the conceptualization of DTs is coined by the
area of its application. Early works on the subject
focused solely on smart manufacturing. A general
understanding of the concept defines a DT as a cyber-
physical system with physical and digital parts. Data
flows between a physical and digital object with full
integration in both directions can be viewed as the
central property of DTs [17]. Second, conceptualization
of DTs needs to be extended by other properties such as
data models, connectivity, accuracy, and
synchronization [13], [24].
Application areas for DTs are expanding and cover
a broad array of domains, from manufacturing and
healthcare to smart cities, logistics, business,
economics, and even psychology. However, a common
reference architecture for DTs can define the relevant
properties of DTs over several domains. Therefore, this
paper presents a unified reference architecture with nine
distinct layers as a blueprint for the configuration of
DTs. Finally, the meta-review unpacks future research
directions for DTs in eight different areas: concept
development, business model, integration, data entry,
big data, data analysis, standardization, and security and
privacy. This opens pathways for future research and
highlights the challenges for the further practical
implementation of DTs.
5. References
[1] A. Fuller, Z. Fan, C. Day, and C. Barlow, “Digital
Twin: Enabling Technologies, Challenges and Open
Research,” IEEE Access, vol. 8, pp. 108952
108971, 2020, doi:
[2] F. Tao and M. Zhang, “Digital Twin Shop-Floor: A
New Shop-Floor Paradigm Towards Smart
Manufacturing,” IEEE Access, vol. 5, pp. 20418
20427, 2017, doi: 10.1109/ACCESS.2017.2756069.
[3] Q. Qi and F. Tao, “Digital Twin and Big Data
Towards Smart Manufacturing and Industry 4.0: 360
Degree Comparison,” IEEE Access, vol. 6, pp.
35853593, 2018, doi:
[4] M. Grieves, “Digital twin: manufacturing excellence
through virtual factory replication,” White Pap., vol.
1, pp. 17, 2014.
[5] X. Wu, V. Goepp, and A. Siadat, “Concept and
engineering development of cyber physical
production systems: a systematic literature review,”
Int. J. Adv. Manuf. Technol., vol. 111, no. 12, pp.
243261, Nov. 2020, doi: 10.1007/s00170-020-
[6] H. Zhang, L. Ma, J. Sun, H. Lin, and M. Thürer,
“Digital twin in services and industrial product
service systems:: Review and analysis,” Procedia
CIRP, vol. 83, pp. 5760, 2019.
[7] K. Y. H. Lim, P. Zheng, and C. H. Chen, “A state-
of-the-art survey of Digital Twin: techniques,
engineering product lifecycle management and
business innovation perspectives,” Journal of
Intelligent Manufacturing, vol. 31, no. 6. Springer,
pp. 13131337, 01-Aug-2020, doi: 10.1007/s10845-
[8] Y. Lu, C. Liu, K. I.-K. Wang, H. Huang, and X. Xu,
“Digital Twin-driven smart manufacturing:
Connotation, reference model, applications and
research issues.,” Robot. Comput. Manuf., vol. 61, p.
N.PAG-N.PAG, Feb. 2020.
[9] N. Mohammadi and J. E. Taylor, “Smart city digital
twins,” in 2017 IEEE Symposium Series on
Computational Intelligence (SSCI), 2017, pp. 15,
doi: 10.1109/SSCI.2017.8285439.
[10] Y. Liu et al., “A Novel Cloud-Based Framework for
the Elderly Healthcare Services Using Digital
Twin,” IEEE Access, vol. 7, pp. 4908849101,
2019, doi: 10.1109/ACCESS.2019.2909828.
[11] R. Parmar, A. Leiponen, and L. D. W. Thomas,
“Building an organizational digital twin.,” Bus.
Horiz., vol. 63, no. 6, pp. 725736, Nov. 2020.
[12] A. Gaggioli, “Digital Twins: An Emerging
Paradigm in Cyberpsychology Research?,”
CyberPsychology, Behav. Soc. Netw., vol. 21, no. 7,
pp. 468469, Jul. 2018.
[13] H. van der Valk, H. Haße, F. Möller, M. Arbter, J.-
L. Henning, and B. Otto, “A Taxonomy of Digital
Twins,” in Proc. 26th Americas Conference on
Information Systems, 2020, pp. 110.
[14] B. Kitchenham and S. Charters, “Guidelines for
Page 4526
performing systematic literature reviews in software
engineering,” EBSE Tech. Rep. EBSE-2007-01,
[15] B. Kitchenham, O. P. Brereton, D. Budgen, M.
Turner, J. Bailey, and S. Linkman, “Systematic
literature reviews in software engineeringa
systematic literature review,” Inf. Softw. Technol.,
vol. 51, no. 1, pp. 715, 2009.
[16] Z. M. Cinar, A. A. Nuhu, Q. Zeeshan, and O.
Korhan, “Digital Twins for Industry 4.0: A
Review,” in Global Joint Conference on Industrial
Engineering and Its Application Areas, 2019, pp.
[17] W. Kritzinger, M. Karner, G. Traar, J. Henjes, and
W. Sihn, “Digital Twin in manufacturing: A
categorical literature review and classification,”
IFAC-PapersOnLine, vol. 51, no. 11, pp. 1016
1022, 2018.
[18] F. Tao, J. Cheng, Q. Qi, M. Zhang, H. Zhang, and F.
Sui, “Digital twin-driven product design,
manufacturing and service with big data,” Int. J.
Adv. Manuf. Technol., vol. 94, no. 912, pp. 3563
3576, 2018, doi: 10.1007/s00170-017-0233-1.
[19] T. Catarci, D. Firmani, F. Leotta, F. Mandreoli, M.
Mecella, and F. Sapio, “A Conceptual Architecture
and Model for Smart Manufacturing Relying on
Service-Based Digital Twins,” in 2019 IEEE
International Conference on Web Services (ICWS),
2019, pp. 229236, doi: 10.1109/ICWS.2019.00047.
[20] K. Y. H. Lim, P. Zheng, and C.-H. Chen, “A state-
of-the-art survey of Digital Twin: techniques,
engineering product lifecycle management and
business innovation perspectives.,” J. Intell. Manuf.,
vol. 31, no. 6, pp. 13131337, Aug. 2020.
[21] T. Y. Melesse, V. Di Pasquale, and S. Riemma,
“Digital twin models in industrial operations: A
systematic literature review,” Procedia Manuf., vol.
42, pp. 267272, 2020.
[22] D. Jones, C. Snider, A. Nassehi, J. Yon, and B.
Hicks, “Characterising the Digital Twin: A
systematic literature review,” CIRP J. Manuf. Sci.
Technol., vol. 29, pp. 3652, 2020.
[23] M. R. Enders and N. Hoßbach, “Dimensions of
digital twin applications-a literature review,” 2019.
[24] C. H. dos Santos, J. A. B. Montevechi, J. A. de
Queiroz, R. de Carvalho Miranda, and F. Leal,
“Decision support in productive processes through
DES and ABS in the Digital Twin era: a systematic
literature review,” International Journal of
Production Research. Taylor and Francis Ltd., 2021,
doi: 10.1080/00207543.2021.1898691.
[25] M. M. Rathore, S. A. Shah, D. Shukla, E. Bentafat,
and S. Bakiras, “The Role of AI, Machine Learning,
and Big Data in Digital Twinning: A Systematic
Literature Review, Challenges, and Opportunities,”
IEEE Access, vol. 9, pp. 3203032052, 2021.
[26] B. Ketzler, V. Naserentin, F. Latino, C. Zangelidis,
L. Thuvander, and A. Logg, “Digital Twins for
Cities: A State of the Art Review,” Built Environ.,
vol. 46, no. 4, pp. 547573, 2020.
[27] M. Perno, L. Hvam, and A. Haug, “Enablers and
Barriers to the Implementation of Digital Twins in
the Process Industry: A Systematic Literature
Review,” in 2020 IEEE International Conference on
Industrial Engineering and Engineering
Management (IEEM), 2020, pp. 959964, doi:
[28] K. Y. H. Lim, P. Zheng, and C.-H. Chen, “A state-
of-the-art survey of Digital Twin: techniques,
engineering product lifecycle management and
business innovation perspectives,” J. Intell. Manuf.,
pp. 125, 2019.
[29] R. Soderberg, K. Warmefjord, J. S. Carlson, and L.
Lindkvist, “Toward a Digital Twin for real-time
geometry assurance in individualized production,”
Cirp Ann. Technol., vol. 66, no. 1, pp. 137140,
2017, doi: 10.1016/j.cirp.2017.04.038.
[30] B. A. Talkhestani, N. Jazdi, W. Schloegl, and M.
Weyrich, “Consistency check to synchronize the
Digital Twin of manufacturing automation based on
anchor points,” in 51st Cirp Conference on
Manufacturing Systems, vol. 72, L. Wang, Ed. 2018,
pp. 159164.
[31] J. Thönes, “Microservices,” IEEE Softw., vol. 32,
no. 1, p. 116, 2015.
[32] A. Rojko, “Industry 4.0 concept: Background and
overview.,” Int. J. Interact. Mob. Technol., vol. 11,
no. 5, 2017.
[33] F. Adrodegari, N. Saccani, C. Kowalkowski, and J.
Vilo, “PSS business model conceptualization and
application,” Prod. Plan. Control, vol. 28, no. 15,
pp. 12511263, 2017.
[34] M. Ghobakhloo, “The future of manufacturing
industry: a strategic roadmap toward Industry 4.0,”
J. Manuf. Technol. Manag., 2018.
Page 4527
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
The use of Digital Twins has gained attraction in research and practice in recent years. Digital Twins are virtual representations of physical objects and they can be connected with their physical counterparts. Through this connection, Digital Twins contribute to the convergence of the real and the virtual world. While existing literature reviews focus strongly on the manufacturing industry, this paper analyzes Digital Twin applications across industries. Based on a systematic literature review, this paper examines 87 Digital Twin applications and proposes a classification scheme with six dimensions to describe the applications identified. The concept of Digital Twins is currently still underrepresented in Information Systems research, which opens up further research opportunities.
Full-text available
Digital twinning is one of the top ten technology trends in the last couple of years, due to its high applicability in the industrial sector. The integration of big data analytics and artificial intelligence/machine learning (AI-ML) techniques with digital twinning, further enriches its significance and research potential with new opportunities and unique challenges. To date, a number of scientific models have been designed and implemented related to this evolving topic. However, there is no systematic review of digital twinning, particularly focusing on the role of AI-ML and big data, to guide the academia and industry towards future developments. Therefore, this article emphasizes the role of big data and AI-ML in the creation of digital twins (DTs) or DT-based systems for various industrial applications, by highlighting the current state-of-the-art deployments. We performed a systematic review on top of multidisciplinary electronic bibliographic databases, in addition to existing patents in the field. Also, we identified development-tools that can facilitate various levels of the digital twinning. Further, we designed a big data driven and AI-enriched reference architecture that leads developers to a complete DT-enabled system. Finally, we highlighted the research potential of AI-ML for digital twinning by unveiling challenges and current opportunities.
Full-text available
Cyber Physical Systems (CPSs) play a crucial role in the Industry 4.0 paradigm. The application of CPSs in production and manufacturing environments gave rise to the term Cyber Physical Production Systems (CPPSs). There is a growing interest in CPPSs, yet research in this area is scattered and needs to be reviewed for understanding their development status and maturity. The aim of this study is to carry out a systematic literature review (SLR) to analyze the current research activities on CPPSs according to their contributions to the engineering life cycle of such production system. Firstly, a method for SLR is presented. Then, literature analysis of CPPSs is conducted to present research activities in the light of the concept development and engineering development stages. Finally, based on the results of the literature analysis, a concept map of CPPSs research is proposed, which depicts the existing research topics in the engineering life cycle of CPPSs. And we exploit it to propose a research agenda of the CPPSs integration process required to ensure their efficient industrial use. Findings of this review can help researchers to examine the maturity of the development status of CPPSs, to discover which phases require improvement, and to know the future research directions for their industrial practices.
Full-text available
Digital Twin technology is an emerging concept that has become the centre of attention for industry and, in more recent years, academia. The advancements in industry 4.0 concepts have facilitated its growth, particularly in the manufacturing industry. The Digital Twin is defined extensively but is best described as the effortless integration of data between a physical and virtual machine in either direction. The challenges, applications, and enabling technologies for Artificial Intelligence, Internet of Things (IoT) and Digital Twins are presented. A review of publications relating to Digital Twins is performed, producing a categorical review of recent papers. The review has categorised them by research areas: manufacturing, healthcare and smart cities, discussing a range of papers that reflect these areas and the current state of research. The paper provides an assessment of the enabling technologies, challenges and open research for Digital Twins.
Conference Paper
Full-text available
Since the last few years, the topic of Digital Twins receives considerable attention in research and with practitioners. A Digital Twin describes the connection between physical and virtual objects. However, a unified definition for Digital Twins is still missing. While existing literature reviews mainly focus on the application of Digital Twins, this review is analyzing the explanations, definitions, and elements of the Digital Twins described in the literature. Based on an examination of 233 papers, this paper proposes a multi-dimensional taxonomy of a Digital Twin and sorts the most common definitions. The taxonomy developed in this contribution allows classifying the various definitions and concepts of Digital Twins that emerged in literature over the years. That may help the reader to gain profound insights into the domain of Digital Twins.
Full-text available
A Digital Twin is one of the enabling technologies of Industry 4.0 that couples actual physical systems with corresponding virtual representation. Currently, the application of Digital Twin models has attracted the attention of many researchers with the focus of production, predictive maintenance, and after-sale services. However, its role in industrial operations particularly in production, predictive maintenance, and after-sales services lacks efforts to systematically review the state-of-the-art. Moreover, this review discusses some of the challenges in implementing DT models to extend its role in the aforementioned application domains. In this paper, a systematic literature review was conducted to assess the role of Digital Twin models in industrial operations and to identify challenges for realization. Twenty-five research studies that were published until the end of June 2019 were selected and analyzed in order to show the current state-of-the-art on the role of Digital Twin models in the industrial operations and challenges in the implementation. Review results underline that the majority of the studies have focused on the application of Digital Twins in the production sector followed by predictive maintenance and after-sales services. Many authors have discussed how to apply Digital Twin models without remarking their role in the aforementioned domains of industrial operations. This paper provides insights for different industrial sectors, practitioners, researchers and experts of the field on the specific roles of Digital Twin models and challenges of implementing these models in the areas of production, predictive maintenance, and after-sale services.
Full-text available
While there has been a recent growth of interest in the Digital Twin, a variety of definitions employed across industry and academia remain. There is a need to consolidate research such to maintain a common understanding of the topic and ensure future research efforts are to be based on solid foundations. Through a systematic literature review and a thematic analysis of 92 Digital Twin publications from the last ten years, this paper provides a characterisation of the Digital Twin, identification of gaps in knowledge, and required areas of future research. In characterising the Digital Twin, the state of the concept, key terminology, and associated processes are identified, discussed, and consolidated to produce 13 characteristics (Physical Entity/Twin; Virtual Entity/Twin; Physical Environment; Virtual Environment; State; Realisation; Metrology; Twinning; Twinning Rate; Physical-to-Virtual Connection/Twinning; Virtual-to-Physical Connection/Twinning; Physical Processes; and Virtual Processes) and a complete framework of the Digital Twin and its process of operation. Following this characterisation, seven knowledge gaps and topics for future research focus are identified: Perceived Benefits; Digital Twin across the Product Life-Cycle; Use-Cases; Technical Implementations; Levels of Fidelity; Data Ownership; and Integration between Virtual Entities; each of which are required to realise the Digital Twin.
The use of simulation to support decision-making in productive processes (goods and services) is already an established research field. However, with the availability of solutions and technologies, simulation is no longer a tool with limited scope and analysis. In this case, the integration of simulation with physical systems is considered to allow virtual models to be sensitive to physical changes and aligned with the current state of processes, forming the so-called Digital Twin. Therefore, the main purpose of this article is to present a systematic literature review of the use of simulation as Digital Twin to support decision-making. We considered studies published in scientific journals and conference proceedings that include the use of Discrete Event Simulation (DES) and/or Agent-Based Simulation (ABS). Although the Digital Twin concept has appeared in recent years, we noted that its principle has been used for decades when it comes to decision-making through simulation. Moreover, there are still many discussions and uncertainties regarding the simulation model in this research field, such as the degree of autonomy, synchronisation, and connection. These and other key issues are discussed and some research opportunities are highlighted, such as the need for constant model validation and integration between various models.
During the last decades, a variety of digital tools have been developed to support both the planning and management of cities, as well as the inclusion of civic society. Here, the concept of a Digital Twin – which is rapidly emerging throughout many disciplines due to advances in technology, computational capacities and availability of large amounts of data – plays an important role. In short, a digital twin is a living virtual model, a connected digital representation of a physical system and has been a central concept in the manufacturing industry for the past decades. In this article, we review the terminology of digital twins for cities and identify commonalities and relations to the more established term 3D city models. Our findings indicate an increasing use of the term digital twin in academic literature, both in general and in the context of cities and the built environment. We find that while there is as yet no consensus on the exact definition of what constitutes a digital twin, it is increasingly being used to describe something that is more than a 3D city model (including, e.g. semantic data, real-time sensor data, physical models, and simulations). At the same time, the term has not yet replaced the term 3D city model as the most dominant term in the 3D GIS domain. By looking at grey literature we discuss how digital twins for cities are implemented in practice and present examples of digital twins in a global perspective. Further, we discuss some of the application areas and potential challenges for future development and implementation of digital twins for cities. We conclude that there are significant opportunities for up-scaling digital twins, with the potential to bring benefits to the city and its citizens and clients.