POLITECNICO DI TORINO

DEPARTMENT OF MANAGEMENT AND PRODUCTION ENGINEERING

MASTER'S DEGREE IN

ENGINEERING AND MANAGEMENT

Digital Twin adoption in manufacturing

systems

Supervisor Candidate

Prof. Alfieri Arianna Carolina Perini

Co-Supervisors

Prof. Pastore Erica

Prof. Castiglione Claudio

Academic year 2023/2024

Table of contents

Acronyms .............................................................................................................................................. 4

List of Figures ...................................................................................................................................... 6

List of tables ......................................................................................................................................... 8

Abstract .................................................................................................................................................. 9

1 Introduction ........................................................................................................................... 10

2 Research methodology ...................................................................................................... 11

3 Introduction to DT technology ....................................................................................... 13

3.1 DTs and Industry 4.0 ...................................................................................................... 13

3.2 The history of DT ............................................................................................................. 15

3.3 Digital Twin trends ......................................................................................................... 17

4 Digital Twin technology: definition and classification in literature ................. 19

4.1 Definitions of Digital Twin ........................................................................................... 19

4.2 DTs classification in literature.................................................................................... 22

5 DTs in manufacturing ......................................................................................................... 25

5.1 The proposed classification framework ................................................................. 25

5.2 General overview ............................................................................................................ 30

5.3 DTs developed frameworks in literature ............................................................... 34

5.3.1 Monitor and improve production process ................................................... 36

5.3.2 DTs for layout design: analysis for optimization ....................................... 41

5.3.3 DTs to handle flexibility ...................................................................................... 43

5.3.4 DTs collaboration ................................................................................................... 48

5.3.5 Cognitive DTs ........................................................................................................... 51

5.3.6 DTs and sustainability.......................................................................................... 55

2
 5.4 Enablers, challenges and opportunities ................................................................. 58

5.4.1 Enablers for DT implementation ..................................................................... 59

5.4.2 Opportunities deriving from DT utilization ................................................ 63

5.4.3 DT implementation challenges......................................................................... 64

6 Conclusions ............................................................................................................................ 68

Bibliography ...................................................................................................................................... 69

3
Acronyms
ADT: Adaptive Digital Twin

AGV: Automated Guided Vehicle

AI: Artificial Intelligence

AR: Augmented Reality

ASC: Autonomous Sub-Component

CDT: Cognitive Digital Twin

CSC: Constrained Sub-Component

CPS: Cyber Physical Systems

DES: Discrete Event Simulation

DSC: Dynamic Sub-Component

DS: Digital Shadow

DST: Digital Sub-Twins

DTMU: Digital Twin Manufacturing Unit

DTP: Digital Twin Prototype

DT: Digital Twin

DTI: Digital Twin Instance

DTMC: Digital Twin Manufacturing Cell

FE: Functional Element

IaaS: Infrastructure as a Service

IoT: Internet of Things

KG: Knowledge Graph

OME: Observable Manufacturing Element

PaaS: Platform as a Service

SoS: System of System

4
UN SDGs: United Nations Sustainable Development Goals

VR: Virtual Reality

5
List of Figures
Figure 1: research methodology ............................................................................................... 11

Figure 2: distribution of the reviewed publications per year ........................................ 12

Figure 3: distribution of the reviewed publication per type of document ............... 13

Figure 4: the four Industrial Revolutions [1] ....................................................................... 14

Figure 5 : Annual investments in digital factory transformation 2022 in billion

US$ (Statista, 2023) ............................................................................................................................... 15

Figure 6: the Mirrored Space Model of Micheal Grieves [6] ........................................... 16

Figure 7: DT’s timeline [7] ........................................................................................................... 17

Figure 8: Interest in the search term 'Digital Twins' in the last 5 years [9] ............. 17

Figure 9: DT market size, by region (USD Million) [10] ................................................... 18

Figure 10: global DT market share in 2020 [11] ................................................................. 18

Figure 11: global DT market size in the year 2020 and 2025, by industry (in billion
U.S dollars) [12] ....................................................................................................................................... 19

Figure 12: integration level of DTs [25] .................................................................................. 23

Figure 13: classification of DT at hierarchical leve [27] ................................................... 24

Figure 14: first classification of the literature ..................................................................... 26

Figure 15: sub-classification of the literature for documents presenting a DT

framework ................................................................................................................................................. 27

Figure 16: Stark’s DT characterization [32].......................................................................... 30

Figure 17: Five dimension DT model [33] ............................................................................. 31

Figure 18 : Redelinghuys’ six layers DT architecture [34] .............................................. 32

Figure 19: DT of a manufacturing process [36]................................................................... 33

Figure 20 : the lifecycle of a DT [37] ........................................................................................ 34

Figure 21 : the process to follow to get started with a DT [36] ..................................... 35

Figure 22: product development process [49] .................................................................... 36

6
 Figure 23: DT framework for geometry assurance [47] .................................................. 38

Figure 24: DT Manufacturing Cell [40] ................................................................................... 39

Figure 25: DT framework for bottleneck identification and throughput

improvement [42]................................................................................................................................... 40

Figure 26: The optimal route planning [43] ......................................................................... 41

Figure 27: Inputs for an automatic generation of a DT [52] ........................................... 43

Figure 28: DT architecture for dynamic scheduling [55] ................................................ 45

Figure 29: DT dynamic scheduling process [56]................................................................. 46

Figure 30: DT general architecture to handle flexibility [58] ........................................ 47

Figure 31: DT architecture for resource planning [59] .................................................... 48

Figure 32: DDTMS [63] ................................................................................................................. 49

Figure 33: Digital Thread-driven manufacturing process [63] ..................................... 50

Figure 34: Framework for automatic error detection of collaborative DTs [64].... 51

Figure 35: Graph learning framework [68] ........................................................................... 53

Figure 36 : Tools supporting cognition [66] ......................................................................... 54

Figure 37: Interconnection of CDTs in the supply chain context [67] ........................ 55

Figure 38: Degree to which consumers' purchasing behaviour and choices shifted
towards buying more sustainable products over the past five years worldwide in 2022
[94] ............................................................................................................................................................... 56

Figure 39: DT contribution to sustainability related functions [70] ........................... 56

7
List of tables
Table 1 : Definitions of DT in literature .................................................................................. 20

Table 2: classification criteria .................................................................................................... 29

Table 3 : DT enablers ..................................................................................................................... 60

Table 4 : DT implementation challenges ................................................................................ 65

8
Abstract
This thesis investigates the role of Digital Twins in enhancing manufacturing
systems within the context of Industry 4.0. The study provides an overview of Industry
4.0 highlighting the importance of digital transformation in modern industrial
practices. The research delves into the concept of Digital Twins, their applications, and
the benefits they offer in terms of optimizing operations, improving efficiency, and
driving innovation in manufacturing industries. Through a literature review, the study
examines the challenges and enablers of Digital Twin adoption. The thesis aims to
contribute to the ongoing discourse on digital transformation in the manufacturing
sector, providing insights for companies seeking to leverage Digital Twins to enhance
their operations and competitiveness.

Questa tesi indaga il ruolo dei Digital Twins nel miglioramento dei sistemi di
produzione nel contesto dell'Industria 4.0. Lo studio fornisce una panoramica
dell'Industria 4.0 evidenziando l'importanza della trasformazione digitale. La ricerca
approfondisce il concetto dei Digital Twins, le sue applicazioni e i vantaggi che offre in
termini di ottimizzazione delle operazioni, miglioramento dell'efficienza e spinta
all'innovazione nelle aziende manifatturiere. Attraverso una revisione della letteratura,
lo studio esamina le sfide e i fattori abilitanti dell'adozione del Digital Twin. La tesi si
propone di contribuire al discorso in corso sulla trasformazione digitale nel settore
manifatturiero, fornendo spunti per le aziende che cercano di sfruttare i Digital Twin
per migliorare le loro operazioni e la loro competitivita.

Keywords: Digital Twin, Industry 4.0, Manufacturing, IoT, Artificial Intelligence,

Simulation, Optimization, Sustainability.

9
1 Introduction
Industry 4.0 has brought a revolution in the manufacturing industry. The term
refers to the fourth industrial revolution that integrates digital and physical
technologies into manufacturing systems. The objective of Industry 4.0 is to create
smart connected factories through the employment of technologies such as Internet of
Things (IoT), Cyber Physical Systems (CPS) and Artificial Intelligence (AI), that enable
real-time collection, analysis, and communications among machines, products, and
people.

The concept of Digital Twin (DT) has gained significant traction in recent years. It
offers manufacturers the ability to create virtual representations of their physical
assets, processes, and systems. DT technology leverages real-time data and advanced
analytics to enable companies to gain valuable insights into the behavior and
performance of their assets. This helps predict maintenance needs, optimize
production processes, and even simulate different scenarios to improve decision-
making. To have a better understanding of what DTs are and how they apply in the
manufacturing sector, a literature review on relevant academic publications has been
conducted.

The objective of this thesis is to investigate the adoption of DTs in manufacturing

systems and the impacts that it may have on efficiency, productivity, and sustainability
manufacturing processes. For this purpose, the thesis is structured as follows: first, the
research methodology used for this thesis is presented, followed by an introduction to
DT technology and its growing trend; then the focus shifts to the general functioning of
DTs followed by their application in the manufacturing context based on the
classification criteria used. The benefits and challenges of using Digital Twins are
finally outlined and the conclusions are presented.

10
2 Research methodology
A Scopus search was conducted to analyse the use of DTs and the benefits derived
from their use in manufacturing industries.

Four types of search were carried out, corresponding to different keywords (Figure
1).

As the aim of the thesis was to analyse the use of DTs in manufacturing companies,
the keywords 'digital twin' and 'manufacturing' were used. Others were added
depending on the objective.

For a general overview, the words 'industry' and 'ecosystem' were added.

In addition, the keywords 'barriers', 'concerns', 'challenges', 'requirements' and

'enablers' were used to analyse the barriers that may be encountered in the
implementation of a DT and the basic requirements that a company should have in
order to successfully exploit this new technology.

Finally, the keyword "sustainability" was introduced to assess whether DTs could
bring benefits in terms of sustainability.

DOMAIN ("Digital Twin" AND "Manufacturing")

KEYWORDS AND
("obstacles"
ADDED ("industry" ("requirements"
OR "concerns" ("sustainability")
KEYWORDS OR "system") OR "enablers")
OR "challenges")

FILTER Relevant papers to the research topic based on: number of citations, contents of
CRITERIA abstract, conclusions and related works

Search #1 Search #2 Search #3 Search #4

TITLE-ABS-KEY

Figure 1: research methodology

Some papers were excluded immediately after reading the abstract, introduction
and conclusion. Others were excluded after a more thorough reading. Finally, 94
documents were selected.

In order to give a general overview of the selected documents, Figure 2 illustrates

11
the number of articles published per year. Looking only at the documents used for this
thesis, we can see that from 2017 onwards the number of publications per year
increases. Although this does not represent the totality of documents related to DTs in
the literature, we can assume from this graph that this year marks a turning point
where the concept of the Digital Twin has matured enough to be discussed more
concretely.

Figure 2: distribution of the reviewed publications per year

The type of articles selected are illustrated in Figure 3: the majority are journal
articles, followed by web page, report and conference proceedings, and finally book

12
sections.

Figure 3: distribution of the reviewed publication per type of document

After introducing DT technology, section 5.1 presents how the documents dealing
with the focus of this thesis (the use of DTs in the context of manufacturing industries)
have been classified.

3 Introduction to DT technology
3.1 DTs and Industry 4.0
The now familiar and established term “Industry 4.0” first appeared in Germany in
2011, in a newspaper article about the German government's high-tech strategy [1].

The term refers to the 4th Industrial Revolution, which is seen as an extension of
the first three industrial revolutions. While these are seen as the result of
mechanisation and the introduction of electricity and information technology, the 4th
Industrial Revolution has been ushered by Internet of Things (IoT) and Cyber Physical
Systems (CPS) [2].

The first industrial revolution began in England in 1780, and then spread to the rest
of Europe and the United States; the mechanised production was born with it, thanks
to a new energy source, coal, and the introduction of the steam engine. Around 1870
the Second Industrial revolution began; new inventions like the electric light, the

13
internal combustion engine, and the first cinema set, were accompanied by a new way
of producing: the mass production. Then, in the second half of the 20th century, we start
talking about the Third Industrial Revolution where traditional industrial technologies
are being abandoned and replaced by digital and computer technologies which allowed
manufacturing automation.

In recent years, we have finally been able to take another step forward. With
Industry 4.0, the digital and physical worlds are converging, thanks to the emergence
of digital manufacturing, network communications, computer and automation
technologies (Figure 4) [1].

Figure 4: the four Industrial Revolutions [1]

The ongoing fourth industrial revolution is made possible by modern advanced

technologies such as autonomous robots, IoT, Augmented Reality (AR), Artificial
Intelligence (AI), and Digital Twins (DTs) [1].

DTs and Industry 4.0 are closely linked as DTs play a vital role in fulfilling various
requirements of Industry 4.0. They are virtual representations of physical objects,
processes, factories, supply networks, and manufacturing lines; they enable real-time
monitoring, predictive maintenance, and data-driven decision-making, all of which are
essential components of Industry 4.0. Furthermore, they provide intelligence to
networked machines on the shop floor, allowing them to organize and execute
production efficiently, which aligns with the goals of Industry 4.0 [3].

On the other hand, DTs make use of the fundamentals of Industry 4.0: technologies

14
such as Big Data, simulation, the Internet of Things and the cloud are just some of the
components required for DTs to operate effectively and provide new functionality to
the business organisation [4].

In summary, DTs and Industry 4.0 are connected through their shared focus on
utilizing advanced technologies, data analytics, and virtual representations to enhance
industrial processes, increase efficiency, and promote innovation in manufacturing and
production.

As will be discussed in more detail below, the integration of the Digital Twin
emerges as a technologically complex process. However, the steady increase in
investments in the context of digitisation attests to the willingness of companies to
undergo a significant transformation. In a report published by Statista [5], we can
observe not only the substantial investments being made in digital transformation but
also how these are not limited to a single sector (Figure 5).

Figure 5 : Annual investments in digital factory transformation 2022 in billion US$ (Statista, 2023)

3.2 The history of DT

Using a general definition, we can say that DTs are a virtual replica of an object, a
production line, a manufacturing process, or a supply chain, that utilises real-time data
to predict the future performance of a machine, a process, etc. [5].

Although the concept of DTs has gained traction in recent years, its origins can be
traced back several years [6]:

Since 1970, the use of simulation and reflective objects has been discussed. NASA
utilised a mirrored replica of inaccessible systems to carry out simulations and find

15
solutions to problems. After an oxygen tank exploded in the Apollo 13 mission,
various simulations were conducted on a replica of the original tank to find a solution
to communicate to the astronauts in space.

In 1991, David Gelernter introduced the concept of a DT in his book "Mirror

Worlds": a software was capable of reproducing the physical world based on specific
inputs [7].

In 2002, Michael Grieves introduced the first DT model, known as the "Mirrored
Space Model". As can be seen from Figure 6, this model included [6]:

• A physical space containing the physical object.

• A virtual space containing the digital replica of the physical object.

The convergence between the physical and digital worlds was ensured by a flow
of data from the physical space to the virtual space and a flow of information in the
opposite direction.

Figure 6: the Mirrored Space Model of Micheal Grieves [6]

In 2006, Grieves' model was renamed into "Information Mirroring Model"; this
highlighted the possibility of representing a single physical reality in multiple virtual
models to explore different alternatives.

Although the concept of a DT already began to emerge, due to the inadequate

technologies available, it was not possible to implement it until 2010 (Figure 7).

In this year NASA published the technological roadmap where, for the first time
ever, the term "Digital Twin" was coined. Here, it was described as:

“An integrated multi-physics, multi-scale, probabilistic simulation of a vehicle or

16
system that uses the best available physical models, sensor updates, fleet history, etc., to
mirror the life of its flying twin” [7].

Figure 7: DT’s timeline [7]

3.3 Digital Twin trends

Although, as mentioned above, the concept of DT first appeared in 2010, it is only
in recent years that it has begun to spread and become increasingly popular.

In 2019, Gartner placed it fourth in its ranking of the 'Top 10 strategic technology
trends for 2019”, ahead of autonomous things, augmented analytics, and AI-driven
development, followed by empowered edge, immersive experiences, blockchain, and
smart spaces [8].

The growing interest in this new technology is evident; indeed, Google Trends
shows us how online searches for 'Digital Twins' increased almost linearly in the last 5
years, peaking in November 2022 and September 2023 (Figure 8) [9].

Figure 8: Interest in the search term 'Digital Twins' in the last 5 years [9]

17
 As we can see from Figure 9, the rise in popularity of DTs has been followed by an
increase in their use on almost every continent. North America leads the way, followed
by Europe, Asia, Latin America, the Middle East and Africa. The highest growth rate is
seen in the top 3, while Latin America and MEA show significantly lower growth [10].

Figure 9: DT market size, by region (USD Million) [10]

Manufacturing industry accounted for over 22% of the global DT market share in
2020, followed by the automotive industry with over 18% (Figure 10) [11].

Figure 10: global DT market share in 2020 [11]

Looking ahead to 2025, Statista [12] predicts that the market value will soar. It will
be the manufacturing sector, which as mentioned above already holds the leadership in

18
terms of market share, that will have the highest growth rate (Figure 11).

Figure 11: global DT market size in the year 2020 and 2025, by industry (in billion U.S dollars) [12]

To summarize, the interest and utilization of DTs is expected to increase

significantly. The manufacturing industry, which has already adopted DT on a wide
scale, will lead the charge in specializing in this new technology more quickly than
other industries. It's not surprising that the most developed countries will invest
heavily in the development of DT.

4 Digital Twin technology: definition

and classification in literature
4.1 Definitions of Digital Twin
Nowadays, many definitions of a DT can be found in literature; Barricelli, Casiraghi,
e Fogli (2019) has grouped them together, and essentially, they can be summarized in
Table 1, where it has been also specified the sector in which the author has applied the
definition.

19
 Table 1 : Definitions of DT in literature

Key points # Definition Sector

Integrated [13] Integrated multi-physics, multiscale, and Aerospace

system probabilistic simulation composed of physical
product, virtual product, data, services and
connection between them

[14] An ultra-realistic integrated multi-physics, Aerospace

multiscale, probabilistic simulation of a system

[15] A systematic approach consisting of sensing, Manufacturing

storage, synchronization, synthesis and service

Clone, [16] Computerised clones of physical assets Manufacturing

counterpart
[17] The virtual and computerised counterpart of a Manufacturing
physical system

Simulation, [18] A safe environment in which you can test the Healthcare
test, impact of potential change on the performance
prediction of a system

[19] Virtual model of physical objects to simulate Manufacturing

their behaviours

Virtual, [20] A virtual representation of the system Aerospace

mirror,
[21] Digital mirror of the physical world Manufacturing
replica

[22] A cyber copy of a physical system Manufacturing

[23] A virtual model of physical object Manufacturing

20
 Both of the first two definitions, formulated by authors who have applied them to
the aerospace sector, highlight several key aspects of its capabilities and characteristics
[13], [14]:

• Integration: DT technology integrates multiple aspects of a complex product,

including its physical and virtual components, as well as the data linking them.
• Simulation: DT technology uses advanced simulation techniques to create a
virtual model of a complex product that accurately reflects its real-life
behaviour.
• Multi-physical and multi-scale: DT technology can simulate a wide range of
physical phenomena and scales, allowing complex products to be modelled with
high accuracy.
• Probabilistic: DT technology incorporates probabilistic modelling techniques to
account for uncertainties and variations in the behaviour of complex products.

On the other hand, Lee, Lapira and Balgheri (2013) [15] extend the context of the
DT. They emphasise the incorporation of data-driven analytical algorithms and physical
knowledge to simulate and monitor the health of a manufacturing machine throughout
its operating life.

Both definitions [16] and [17] refer to DTs in the context of manufacturing,
highlighting their virtual nature. However, while in the former the focus is on the fact
that the DT constitutes an exact duplication, thanks to the use of the term 'clone', in the
latter, with the use of the term 'counterpart', a parallel existence is implied,
emphasising the DT as an entity corresponding to the physical system.

Definitions [18] and [19] of DT focus on the purpose and usefulness of this
technology in the areas of testing and simulation. The difference between the two
definitions is subtle. In definition 8, the DT is viewed as a tool for simulating
hypothetical situations or "what if" scenarios. On the other hand, Definition 9 defines
it as a tool that anticipates the behavior of the corresponding physical object.

In relation to the final section of the table, while all definitions agree on the virtual
representation of physical systems or objects, each author introduces a distinct nuance.
Definition [20] has a broad emphasis on the virtual representation of the system.
Definition [21] highlights the reflection of the physical world through the concept of a

21
digital mirror. Definition [22] emphasises the cybernetic nature and duplication of the
system. Finally, Definition [23] focuses on virtual models of physical objects. These
differences highlight the conceptual richness and interpretative aspects of the DT.

The diversity of definitions proposed to describe a DT reflects its versatility and the
breadth of possible applications. It is clear that there is no single universally accepted
definition, but rather a set of perspectives that fit specific contexts and objectives.

4.2 DTs classification in literature

After introducing the concept of DT and providing various definitions, it might be
interesting to analyse how they have been classified in the literature before presenting
the classification criteria for this thesis.

Various documents present distinct classification criteria based on their respective

objectives. To offer a comprehensive and diverse outlook on these criteria, we have
collected a range of them.

Grieves M. and Vickers J. (2016) [24] categorized the different types of DTs on the
base of its creation time in relation to its physical counterpart, identifying two types of
DTs: Digital Twin Prototype (DTP) and Digital Twin Instance (DTI). The first one is
created before the physical object, allowing for aesthetic and functional tweaks to be
made before the production begins, while the latter is created after its physical twin
has been produced and remains "attached" to it throughout its lifecycle.

Kritzinger et al. (2018) [25] identified three types of DT based on the level of
integration between the physical and digital object. The differences among these types
are determined by the data exchange that takes place (or not) between the physical and
digital world (Figure 12):

1. Digital Model: in a digital model, there is no exchange of data between the

physical product and its DT. Data from the physical product is used to
construct the digital product, but if the physical product were to change, the
DT would not automatically change.
2. Digital Shadow: in this case there is a one-way exchange of data so that a
change in the physical object would result in a corresponding change in the
digital one.

22
 3. Digital Twin: in this case a two-way flow of data takes place. Whereby any
change, whether occurring in digital or physical space, would be reflected in
the corresponding physical/digital object.

Figure 12: integration level of DTs [25]

Another type of classification that has been found in literature is relative to the
intended use of a DT, so that the following types can be identified [26]:

• Product Digital Twins: it is used to represent a physical product in virtual

space. The DT of a product makes it possible to simulate different
conditions, see how the product reacts and then make changes to the
design until the ideal one is achieved.
• Production Digital Twin: this type of twin makes it possible to assess the
functioning of an entire production process. By simulating production,
they can reveal possible inefficiencies in the production line (such as
bottlenecks) before they manifest themselves in the physical world. By
combining the DT of machinery (DT of a product) with that of production,
we can also predict when maintenance will be required.
• Performance Digital Twin: these twins are able to harness the data
produced by smart assets and products, then aggregate, analyse and
enable informed decision making.

According to Tao et al. (2019) from a hierarchical perspective – i.e the level of
involvement within the company - DT can be divided into three different levels (Figure
13) [27]:

23
 • Unit level: this level refers to the smallest participant in the production
process. It can be a material, a component or a piece of equipment.
• System level: multiple DTs at the level of units exchanging data form a DT
at the system level. At the system level we can have a production line, a
department or an entire factory.
• System of Systems (SoS) level: several DTs at the level of units
exchanging data form a DT at the system level. This could be a production
line, a department or an entire factory. Sometimes a complex product can
be represented by such DTs, which can also evaluate and improve the
interaction between multiple components.

Figure 13: classification of DT at hierarchical leve [27]

Madni et al. (2019) identifies four levels of virtual representation based on the
sophistication level [28]:

• Pre-Digital Twin: it is built before a product is prototyped to help guide

design decisions and identify problems early on.
• Digital Twin: it represents the DT as we have introduced it so far.
• Adaptive Digital Twin: it can support real-time planning and decision-
making during operations, maintenance, and support by acquiring the
preferences of the operator or user.
• Intelligent Digital Twin: in addition to all the characteristics of the
previous levels, this type of DT also has learning capabilities.

24
 Other rwo classification criteria are the ones identified by Yu et al. (2022) and
Enders et al. (2019). The first one classified the DTs on the base of their application
scale: nano (molecular level), Micro (single operation or part), Meso (collection of
operations) and Macro (community, local area) [29]. The second one instead divided
the DTs depending on the purposes: simulation, monitoring and control [30].

5 DTs in manufacturing
5.1 The proposed classification framework
After having reviewed various DTs classification criteria in the literature, this thesis
will adopt a specific criterion that will be used throughout the following sections.

Looking at the documents related to the adoption of DTs in manufacturing

companies, this being the focus area of the thesis, it is possible to make an initial
distinction between three main categories of approaches (Figure 14).

The first category includes documents that provide an overview of the general
operation of DTs, focusing on explaining the basic principles and their overall
functioning.

The second identified category includes documents presenting specific DT

frameworks designed for particular purposes within manufacturing companies. These
documents focus on the structure and application of DT models.

Finally, the third category includes documents that do not deal directly with the
development or application of DTs, but rather provide guidelines regarding the
enabling factors, challenges and opportunities of using a DT within manufacturing
frameworks.

25
 Figure 14: first classification of the literature

For the documents presenting a general overview no further classification has been
adopted.

For the second type of documents, those presenting a DT model, a further

categorisation criterion was introduced. In particular, the purpose for which the
presented DT was developed was considered. More specifically, we have identified 6
different functions:

• Monitor and improve the production process: use the DT to monitor and
optimise the production process, allowing more efficient management and early
identification of potential improvements.
• Design the layout in an optimised way: use the DT for optimised layout
design of equipment and resources, improving spatial and operational
efficiency.
• Enhance sustainability: employ the DT to monitor and optimise resource
utilisation, reducing environmental impact and promoting sustainable
production practices.
• Handle flexibility of the production system: Leverage the DT to quickly and
efficiently adapt the production system to changes in demand or other changes,
ensuring greater flexibility.
• Collaboration with other DTs: use the DT as a platform for collaboration
between different Digital Transformation systems, facilitating the exchange of
data and information to optimise interconnected processes.
• Cognitive DT: to offer enhanced predictive analysis, decision-making and
optimisation capabilities thanks to its cognitive functionalities.
26
 Moreover, for each function, it has been identified that DTs can offer different
services:

• DT Services: this refers to what the DT does to fulfil its purpose. On the base of
the analysis of the documents found in literature we have identified that a DT
can perform real-time monitoring, support decision-making, predict potential
failures, and conduct optimization analysis.
o Real time state monitoring: the DT allows real-time monitoring of a
system or object, providing immediate data on its current status. This
allows constant and up-to-date observation of the operating conditions.
o Decision-making support: the DT provides detailed and contextual
information to support decision-making. It helps to make informed
decisions based on the data and simulations generated by the digital
model.
o Failure analysis and prediction: by analysing historical and real-time
data, DT is able to identify potential faults and predict future problems.
This functionality allows preventive intervention to avoid malfunctions.
o Analysis for optimization: the DT allows in-depth analysis of processes
and operations, facilitating the identification of areas where efficiency
can be optimised, costs reduced, or overall performance improved.

To summarize, the documents presenting a DT framework has been divided on the

base of the DT purpose. And for each document has been identified the DT service and
the scope (Figure 15).

Figure 15: sub-classification of the literature for documents presenting a DT framework

27
 The documents that fall under the third classification criterion (enablers,
challenges and opportunities) were subclassified. The enabling factors were further
divided based on their dependence on specific technologies, adherence to processes,
essential employee skills, or a well-defined corporate culture. This approach helped in
gaining a more detailed understanding of the factors that contribute to organizational
success. The challenges are divided in 3 building blocks: engineering related
challenges, organizational challenges and data related challenges. Instead, for the
opportunities no further classification has been adopted.

Table 2 summarizes all the classifications that have just been listed. On the left we
can see the first classification (general overview, DT frameworks and enablers,
opportunities , and challenges); the second column shows the subclassification (for
example for the documents showing a DT framework this column illustrates every
function DT can have, like monitor and improve production process); the third block
of the table shows instead the transversal classification (for example, for a specific DT
function, it is illustrated the service that DT offer).

28
 References
OVER
VIEW
[31], [32], [33], [34], [35] [36], [37],
DT services
Function Real time state Decision making Failure analysis and Analysis for
monitoring support prediction optimization
DT FRAMEWORKS

Monitor and improve [38], [39], [40], [41], [44], [45], [46], [47],
production process [42], [43] [48], [49], [50]

Layout design [51], [52], [53], [54]

[55], [56], [57], [58] [55], [56], [58] [59],
Handle Flexibility
[59], [60], [61] [60], [61]
DTs collaboration [62] [63], [64]
Cognitive DTs [65], [66], [67] [68], [67], [69] [66] [65], [68], [66]
Enhance Sustainability [70], [71], [72] [72], [73] [73], [74], [75], [76]
Category
Systems and People and
OPPORTUNITIES &

Process Culture and strategy

Technologies competences
CHALLENGES
ENABLERS,

Enablers [77], [78], [79], [80] [77] [77], [81] [77]

Opportunities [82], [83], [84], [85], [86], [87]
Category
Engineering Organizational Data
Challenges [88], [89], [90], [91], [92],
[89], [90], [92], [93] [89], [90], [91], [93]
[93]
Table 2: classification criteria
5.2 General overview
This section is dedicated to understanding how a DT works, i.e. how the interaction
between the physical world and the real world takes place.

As previously mentioned in subsection 3.2, Michael Grieves provided the initial

representation of how a DT operates in 2014. He depicted it as a connection between
physical and virtual space [31].

Stark et al. (2017) [32] characterises the DT as the combination of an asset's Digital
Master model, its individual Digital Shadow and an intelligent linking of the two. It
involves the digital shadow being created through operation and condition data,
process data, etc., generated by the individual product or production system (Figure
16).

Figure 16: Stark’s DT characterization [32]

Another proposition is made by Tao et al. (2018). Their five-dimension DT model

includes the following components (Figure 17) [33]:

• Physical entity (PE): the actual physical equipment being monitored.

• Virtual entity (VE): the DT or virtual model of the equipment.
• Sensor system (Ss): the system of sensors used to collect data from the
equipment.
 • DT data (DD): the data collected from both the physical and virtual aspects of
the equipment, as well as their fusion.
• Connection model (CN): the bidirectional connections between the PE, VE, Ss,
and DD to facilitate data exchange and analysis.

Figure 17: Five dimension DT model [33]

Similarly, Redelinghuys et al. (2019) [34] propose a six layers DT to illustrate the
data information flow (Figure 18). The first two layers contain the physical twin: the
first one includes all the physical devices, while the second one the local controllers
which provide some functionalities to the DT. Layer 3 contains the data which reflect
the details of the physical twin. Layer 4 acts as gateway between layer 3 and 5, selecting
only the data that need to be transmitted. Layer 5 contains the database servers that
act as repositories of the information transmitted by the gateway in layer 4. Finally,
Layer 6 represents the intelligence of DT: here is where emulation and simulation take
place.

31
 Figure 18 : Redelinghuys’ six layers DT architecture [34]

Boje et al. (2020) [35] have defined DT architecture consisting of three distinct
layers, comprised of a variety of components and technologies. The physical layer
represents tangible entities and reflects the product life cycle stage. The network layer
links the physical and virtual domains and enables the data and information exchange.
The computing layer is comprised of the virtual entities that replicate their physical
counterparts with data-driven and physics-based models, as well as services and users.
Key DT components such as information structures, models, software technologies,
hardware technologies, etc. play a paramount role across these layers [35].

If we compare the frameworks illustrated above, we can see that the application
areas of Digital Twins are many. The frameworks illustrated by Stark [32] is used within
cyber physical system for testing during ongoing operations to ensure error-
management real-time performance and analysis. The DT in [33] is employed to
monitor work conditions of products difficult to accede, with lots of components that
may fall (for example wind turbine). The focus of [34] is instead the exchange of data
and information between a remote simulation and a manufacturing cell. Finally [35]
illustrated a construction DT whose benefit, in the built environment, is the accrual of
knowledge about the physical world delivering improved lifecycle costs and bult asset
resilience.

According to a report by Deloitte [36] , the process can be defined as sensors

32
gathering data about the machines, production line and environment, which is
transmitted to the DT for analysis. Any deviation from the ideal are flagged, and the
production process is changed. Communication interfaces are used, and a number of
security measures including firewalls, encryption and device certificates are needed.
The key components of this process are also explained (Figure 19): Sensors to gather
data; Data that must be compared against company records to find the discrepancies;
Data must be transmitted to the DT where the digital and physical worlds are overlaid;
Data received by the DT, so that it can be modeled; Actuators that can then adjust the
process as needed.

Figure 19: DT of a manufacturing process [36]

Shifting the focus on how the DT works, Colin J. Parris [37] identifies three phases:
see, think, do (Figure 20).

In the seeing phase, the DT collects data to give a warning when a certain threshold
is reached and then predicts the nature of the problem. The model, like the one in
question, can update itself to represent the exact conditions of the physical product,
second by second. In the second phase (thinking phase), the DT provides options for
the user to pursue. To do this, the DT runs simulations looking at historical data, real-
time data, cost and revenue forecasts. Each proposed option is accompanied by an

33
explanation of the risks involved and the confidence level. Finally, in the dew stage the
option selected by the operator is executed. If the one proposed it is a manual operation
that must be done by an operator, then the operator himself is informed of the
conditions required for the operation [37].

Figure 20 : the lifecycle of a DT [37]

5.3 DTs developed frameworks in literature

The aim of this section is to present some DT frameworks - i.e. a conceptual and
technical structure designed to create and manage a digital twin - in the context of
manufacturing companies, focusing on specific DT functions.

Before examining the different frameworks found in literature, it is important to

understand how a company can implement DTs that enable a virtual representation of
physical assets, systems, processes, products, services, or people.

Qamsane et al. (2021) identifies the following steps [44]:

• Planning: it involves determining if there is a need to enhance some aspect of

the manufacturing ecosystem and if that need could be addressed through the
application of a DT solution.
• Requirements and Analysis: it involves studying and analysing the
requirements for the DT design and development activities.

34
 • Design: it defines a design of a DT solution based on the recommended
alternative that will meet functional, data, and interaction requirements.
• Development: it aims to transform the output design stage into a complete
working DT solution that can address the manufacturing needs established in
the planning stage.

Similarly, Deloitte [36] identifies 6 steps (Figure 21). The first one is to imagine all
the possibilities: create a list of all the potential scenarios where it can be applied.

Figure 21 : the process to follow to get started with a DT [36]

Once the list is prepared, it is crucial to identify the pilot product or process that
can provide the most value and where the implementation of DT is likely to succeed.
Through iterative and agile cycles, we can start with the pilot and improve step by step.
The next step is industrialisation: establishing a standardised and structured
approach to the development and implementation of DTs. To scale the twin, the next
step is to identify additional processes or products that are related to the pilot project.

As with any project, it is important to monitor if it is working properly once it has

been completed.

Now we can go deep into the pilot phase (using the terminology of Deloitte article)
or the design phase (using the Qamsane’s terminology). We will show different DTs

35
frameworks accordingly to the classification criteria used in 5.1; so we have divided
the different subsection on the base of our classification criteria, which is based on DT’s
functions: monitoring and improving production processes, layout design, flexibility
management, collaboration between DTs, and cognitive DTs.

5.3.1 Monitor and improve production process

As shown in section 5.1, DT could be used to monitor and improve production
processes by providing decision support and optimisation analysis.

Analysis for optimization

In the context of implementing DTs to improve manufacturing processes, we will

focus on two key aspects: (1) optimising the product development process and (2)
using DTs for zero-defect manufacturing. DTs for product development process
optimisation provide a virtual approach to design and validation, reducing the reliance
on expensive physical prototypes and enabling faster and more efficient iterations[45].
At the same time, DTs for variation management play a key role in ensuring the
consistency and accuracy of geometric specifications, helping to reduce unwanted
variation in final products [46].

The product development cycle has three key phases: product design, product
validation and product manufacturing (Figure 22).

Figure 22: product development process [49]

For product design Ma et al. (2019) [49] presented a framework for digital twin
augmented human-machine interaction during conceptual design phase. In the virtual
36
world designers can interact with the virtual model via AR/VR technologies, which
allow designers to “see”, “touch” or “interact” with the model.

Xiang et al., (2019) [48] proposed a DT based technique to intelligently select green
materials, to identify the most suitable material for a given design; their DT model can
imitate and improve the qualities of materials that are potentially selected, in order to
make a repetitive comparison between forecasted qualities of these materials and
required properties.

Regarding product validation phase, this is still a significant challenge in terms of

cost and time. Traditionally, this phase involves the use of physical prototypes to assess
the feasibility of the previously designed product [47].

Huang et al. (2022) [47] proposed a framework where a virtual replica of the
manufacturing system is created, allowing for the development of a virtual prototype,
which exactly reproduces the physical prototype by replicating the operations that
would have been performed in the physical world; in the virtual space, the validation
phase of the prototype can be detailed, and if the virtual prototype passes successfully
this phase it is considered ready for production, otherwise an iterative process is begun
in which the design is optimised, and re-tested in the virtual world.

While each framework addresses different aspects of product development, they

can be complementary. For instance, the virtual prototype created in the validation
phase by Huang et al. (2022) [47] can benefit from the use of green materials selected
in the design phase using the DT model proposed by Xiang et al. (2019) [48]. Even if
each framework addresses specific aspects of product development, collectively
providing a more holistic understanding and optimization opportunities there may be
challenges in integrating DT frameworks seamlessly, particularly if they rely on
different technologies, data formats, or modeling approaches. Contradictions may arise
if the results or recommendations from one DT framework conflict with those from
another. For example, the selection of certain materials in the design phase may lead to
unforeseen issues during virtual validation, necessitating iterative refinement.

While DTs have revolutionized the product development space and focused largely
on the design, validation, and manufacturing phases, there is yet another area critical
to consider — variation management or geometry assurance during the manufacturing
phase. Now the information available through the DTs for product development from
37
design and validation can be integrated with DTs for variation management, by re-
purposing the simulation model used for the validation phase, to ‘see’ the geometric
variations of the part in real-time as it proceeds through the assembly process. [46].

With this aim Figure 23 represents a DT framework for geometry assurance.

Figure 23: DT framework for geometry assurance [47]

Two parts A and B are scanned, and the data collected is entered into the simulation
model. The data is sorted by class to achieve a pairwise match that minimizes assembly
variation. In addition, the deviation from the standard can be reduced by careful
analysis of how to adjust the weld points (locating schemes) and by defining the
optimal welding sequence to follow [47].

Decision making support through real-time state monitoring

In this section we will explore the potential of DTs in real-time monitoring to

support operational decisions and improve the performance of production processes.
In particular, we will look at how the integration of a DT into a production cell can
enable improved autonomy, how real-time data acquisition can identify and mitigate
bottlenecks within a manufacturing system, and how the use of real-time location
information can optimise production logistics. We will explore how these aspects,
supported by an effective implementation of the DT, can contribute significantly to the
dynamics and efficiency of manufacturing operations.

By integrating a DT into a manufacturing cell, we can examine how this virtual

entity interacts with the production process in real time, capturing detailed data and
providing a constant flow of information. To this end, we can define the DT
Manufacturing Cell (DTMC) as “a minimum implementation unit for industrial

38
enterprises to put intelligent manufacturing into practice” [38].

Thanks to its structure, which consists of five different types of spaces, the cell is
capable of independent decision-making and proposing improvements [39]. More
precisely, the layers that make up this entity are as follows (Figure 24) [40]:

• Physical space: thanks to sensors the status of the physical process, such as
WIP, can be monitored in real time.
• Virtual space: here is where all the data collected in the physical space come
together. Thanks to them, and thanks to DTs of the physical elements involved
in the production process, it is possible to carry out a simulation in virtual space,
which makes it possible to predict and, if necessary, improve.
• Data space: before entering the virtual space, data on WIP, machine status and
other process elements are transported here to be pre-processed.
• Knowledge space: a dynamic knowledge base in this area enables
improvement decisions to be made.
• Social space: it integrates various service systems like CRM and ERP, bridging
the gap between DTMC supply and customer demand.

Figure 24: DT Manufacturing Cell [40]

Within an intelligent production cell, multiple improvements can be made without

the need for human intervention. An example is the diagnosis and improvement of
bottleneck throughput: Mahesh et al. (2023) [42] proposed a framework (Figure 25) in

39
which, in line with what has been said above, data relating to an Observable
Manufacturing Element (OME) - such as a machine, a process or an entire physical
system - is collected and pre-processed. Processing means [42]:

• Cleaning: handling missing instances and standardisation.

• Integration of different types of data from the physical space.
• Transformation of data.

After collecting event logs and timings, a dynamic map of resource dependencies
and interactions is created. The resulting information is then used by the DT to replicate
the OME. The utilisation rate is determined by monitoring individual resources using
asset monitoring. Finally, prescriptive analytics identifies the busiest resources and
targets them for DT improvement opportunities.

Figure 25: DT framework for bottleneck identification and throughput improvement [42]

Real-time data can be used not only in production processes, but also in logistics
operations in order to reduce the high costs linked to them.

Here, we will only discuss one logistic operation that can benefit from
implementing a DT: Automated Guided Vehicles (AGV).

An AGV is an autonomous vehicle designed to move in industrial or logistical

environments, guided by technologies such as sensors, machine vision or magnetic
guidance systems, without the need for human guidance.

40
 The utilisation of a DT enables optimal route planning in a highly dynamic
environment (Figure 26). The AGV can effectively scan the surrounding area and
determine the most favourable path to a set destination via a simulator. Nonetheless, it
is not competent enough to make dynamic decisions on its own. If an unexpected
obstruction emerges on the route, the AGV must come to a halt, rescan the
surroundings, and wait for the simulator to compute a new route. The integration of DT
can significantly reduce the time spent, particularly in a dynamic setting. DTs can gather
real-time data and update information pertaining to the surroundings at set intervals.
Subsequently, a potential barrier can be detected beforehand, allowing the simulator to
calculate the most effective route before the obstacle obstructs the automated guided
vehicle [43].

Figure 26: The optimal route planning [43]

5.3.2 DTs for layout design: analysis for optimization

As to production system and assembly line design, element arrangement is critical
for effectiveness. Layout design issues may be crucial, affecting operational efficiency,
internal logistics and the overall workflow. DTs emerged for this matter as essential
tools, offering the capability of simulating, analysing and optimising the layout in the
pre-physical implementation phase. We will delve into how DTs can help make better
decisions and improve the design of production systems and production lines.

Guo et al. (2021) [53] developed a DT-based layout optimization approach by

41
proposing three sub-frameworks: (i) the workshop partitioning sub-framework, which
optimizes the workshop layout by analyzing twin data and simulating different
partitioning configurations; (ii) the equipment layout optimization sub-framework,
which adjusts the layout of equipment and facilities by collecting real-time data and
value-adding to twin data; and (iii) the distribution route optimization sub-framework,
which optimizes the material distribution route in order to reduce WIP backlog and to
improve tooling efficiency. In this way, by leveraging real-time feedback and data
analytics, the DT-based approach can optimize the layout of the workshop, the
placement of equipment, and the distribution route of materials in order to improve
production efficiency, reduce WIP backlog, and increase tooling utilization.

For the optimization of layout, a simpler model exists where instead of production
line layout, the DT takes, as an input, the positioning of machines in a production line,
layout of production lines, and scheduling of production processes, makes a simulation
and presents outputs. The outputs can be turned into performance metrics such as
productivity, throughput, efficiency, by use of metrics such as machine cycle time,
material handling time, operator travel time. Therefore, the plant layout can be
optimized by adjusting the positioning of machines, processes, and workstations on the
factory floor. The optimized layout configuration is re-simulated to evaluate its
performance and compared to previous configurations [54].

Based on dynamic and real-time changes in production processes, Lee et al. (2022)
[51] suggest a DT framework, to optimally respond to these changes by suggesting
improvements. The framework proposed consists of 2 layers:

• Information layer: this layer, which is integrated with the enterprise resource
planning (ERP) and manufacturing execution systems (MES), contains
information about the manufacturing design, the resources employed and the
manufacturing bill-of-material.
• Application layer: it includes (a) an interface module which links data
collected from the information layer to the simulation and optimization
modules, (b) a DT simulation module to visualize and verify production
processes and (c) an optimization module that uses algorithms to refine the
process configuration and production line layout.

These frameworks can be compatible if integrated carefully. For instance, Guo et

42
al.'s (2021) approach could benefit from the dynamic adjustments suggested by Lee et
al. (2022). Overlaps may occur since both of them involve real-time data collection.
Also, turning the outputs of simulations into performance metrics could be integrated
with the other two models. However, there could be contradictions if more models are
used in the same time if they prioritize different optimization criteria.

One way to transfer data from the interface module to the DT simulation model it
could be used the framework proposed by Sommer et al. (2023) [52], to automate the
DT generation. By scanning the shop floor and comparing the scanned objects with the
objects’ CAD existing in a reference database, a DT can quickly be created, using inputs
like objects parameters (machine geometry and its positioning information),
parameters that we can obtain via object recognition, all organization’s specific
parameters which we can not obtain through scanning (i.e. machine ID) (Figure 27).

Figure 27: Inputs for an automatic generation of a DT [52]

5.3.3 DTs to handle flexibility

Being able to respond quickly to changes in customer needs, market requirements,
and operational situations has become a crucial element for achieving success.
Adaptability is essential for dealing with challenges such as fluctuations in demand,
customized products, and emerging technologies. This flexibility -i.e the production
system’s capability of changing its internal characteristics- leads to better
responsiveness, competitiveness, and long-term sustainability, enabling companies to
navigate through dynamic scenarios with effectiveness.

43
 Analysis for optimization through real-time state monitoring

To handle flexibility the frameworks found in literature all uses real-time state
monitoring in order to make optimization to the manufacturing process.

To address manufacturing flexibility Zhuang et al. (2018) developed a DT

framework for production management in the context of a product assembly shop floor.
Here, real-time data from the physical shop floor flow to its DT which mirror the
physical operation conditions and simulate the future’s production operations.
Dynamic data like temporary assembly tasks added, product lead time changed,
equipment shutdown or failure, product quality problems found, product design and
process changed, and other kinds of production disturbances are taken into
consideration [60].

Similarly, Park et al. (2019) focused on the application of a DT in the context of

personalized production, where production processes for different product groups
struggle to respond to the dynamic situations arising from these processes,
demonstrating that dynamic situations in the personalized production can be
effectively handled through a combination of five applications: digital twin application,
context-aware application, advanced planning application, advanced scheduling
application, and device control application [61] .

To manage flexibility Yan et al. (2022) developed a DT framework for dynamic

scheduling, allowing for real-time responses to changing operational variables and
needs. In operating systems, an important variable to consider is machine failures.
Should these occur or be predicted with enough lead time, rescheduling becomes
necessary. If an anomaly is detected in the physical system, the information is
transferred to the DT. The DT can then use simulation to reschedule production, taking
into account new parameters such as a machine being unavailable (Figure 28) [55].

44
 Figure 28: DT architecture for dynamic scheduling [55]

To achieve optimal re-scheduling Tliba et al. (2023) [56], presented a DT composed

of two connected models within the physical system:

1. A scheduling model
2. The shop floor model with which simulation is carried out.

The initial scheduling is generated by the first model, based on data contained in
the company's ERP concerning available resources, product details, and company
constraints. The second model simulates the provided scheduling by triggering a
simulation loop. If the result of the simulation is not optimal, the data is updated and
reinserted into the scheduling module, which makes a new proposal to the shop floor
model that generates a new simulation. This process continues until an optimal
solution is found (Figure 29).

45
 Figure 29: DT dynamic scheduling process [56]

While the four frameworks illustrated above focus on how overcome internal
disruptions and dealing with internal sources of flexibility, Neto et al. (2023) [58],
focusing on how to deal with customers’ changing demand requirements, developed a
DT framework with the aim of helping manufacturers to deal with mix flexibility -i.e
the ability to change the short-term production mix in order to implement a desired
sales strategy. In their proposed architecture (Figure 30) the shop floor is replicated in
the virtual world through sensors that collect real-time data from the production
system. This data includes information about the machines, buffers, processing times,
routes, production schedules, and the position of pieces within the machines and
buffers. All of this data enables the DT to simulate, returning an estimation of the
production system key performance indicators (throughput), the expected delivery
date for all products, and the predicted time slots in which the machines are expected
to be idle to perform maintenance [58].

46
 Figure 30: DT general architecture to handle flexibility [58]

Dos Santos et al. (2021) [59] also focused on implementing a flexible production
line to meet changing market demand and minimise operational waste caused by
unused operators or excessive production that does not meet the real needs of the
customer. In the framework developed (Figure 31) the first step is the analysis of sales
history using AI to predict future demand. It then uses a Discrete Event Simulation
(DES) model within a DT architecture to simulate the production process, testing
different variables such as the number of operators. Finally, a dashboard shows the
guidelines for operational planning (resulting from the DES model), including the
optimal resource sizing, the expected production (which may be differ from the
expected demand in the case of batch production) and the expected lead time.

47
 Figure 31: DT architecture for resource planning [59]

5.3.4 DTs collaboration

We will now focus on analysing how using multiple DTs which are able to
collaborate can add significant value. It is in this context that the Digital Thread was
introduced [62]:

“A Digital Thread connects real things and their twin models, but also the
communication networks, the decision algorithms, the visualisations needed to work in
design, construction, and operation within a mature Industry 4.0 environment”

Failure analysis and prediction

It is possible for example to assemble several DTs of every production unit to

recreate an entire production line in the virtual world. This can help in the prediction
of possible failures, thus assuring a higher quality of the production process.

Liu et al. (2023) proposes a framework with the goal to ensure the quality of a final
assembly; it can be interpreted as the evolution of the process regarding assembly
through geometry assurance that we have seen in 5.3.1.

Before introducing an architecture for DTs collaboration we must firstly give some
definitions that we will use later on (Figure 32) [63]:
48
 • Digital Twin Manufacturing Unit (DTMU): it is “the unit-level manufacturing
system with a manufacturing function in the workshop and has the essential
characteristics of a DT system” .
• Distributed Digital Twin Manufacturing System (DDTMS): it is “a workshop-
level intelligent manufacturing system. It has three manufacturing spaces,
including the workshop, agent, and manufacturing unit layer” .

Figure 32: DDTMS [63]

• Unit twin layer: it is the set of all DTs of each production unit (assembly and
machining units).
• Mid-Agent layer: it receives the information provided by the Unit Twin layer
to analyse the production process through simulations. It is also the
communication medium of each manufacturing unit.
• Workshop layer: it analyses industrial big data to define manufacturing tasks
and optimizes combinations of manufacturing services.

The Digital Thread runs through the entire life cycle of the manufacturing process
(machining, inspection, assembly, final inspection) (Figure 33). To do that its structure
it is split into two layers:

49
 • The model analysis layer receives information about the quality of each part
to reproduce a DT and simulate all the operations and estimate the quality of
the final product.
• The data network analysis layer analyses the production area at each stage
and, based on the quality defect transfer, is able to adjust the next operation.

Figure 33: Digital Thread-driven manufacturing process [63]

Thanks to this structure, problems can be easily identified, and production adjusted
accordingly.

Since the functioning of the Digital Thread depends on the functioning of several
DTs, Sahal et al. (2021) analyse how it is possible to identify erratic operational data
that can occur from each DT.

50
 Figure 34: Framework for automatic error detection of collaborative DTs [64]

In Figure 34 a high-level framework for automatic error detection of collaborative

DTs is presented. The analysis assesses whether the data from the DT indicates a failure
or inconsistency, determining if the issue is localized or affecting all nearby DTs
representing the same device. If a widespread problem is detected across all deployed
DTs, the system can proactively notify decision-makers about the failure. On the other
hand, if the issue is specific to one DT, the system disregards the inaccurate data,
replacing it with the correct information to maintain consistency in the operational
data across the production system [64].

The Liu’s Digital Thread [63] ensures a comprehensive understanding of the

manufacturing process, while Sahal et al.'s framework [64] enhances data reliability
within this thread. By combining these models, manufacturers can swiftly identify and
rectify issues, ensuring smooth operations and high-quality output.

5.3.5 Cognitive DTs

Cognitive DT (CDT) is an advanced perspective of the traditional DT and signals a
turning point in the evolution towards Industry 4.0. Such a novel approach of CDT
transcends the simple digital translation of physical objects and, in fact, involves the
employment of artificial intelligence capabilities together with advanced data analytics.
Indeed, cognitive functions make it feasible to transfer knowledge that is gained in one

51
field to another field [65].

Skills which constitute the bedrock of cognition - attention, perception, memory,

reasoning, learning, and problem solving. Herewith follow definitions of the terms that
have been readapted to the DT context [68]:

• Perception: that's the process of preparing meaningful representations of the

data relating to the physical twin and its surrounding physical environment for
further processing.
• Memory: recall data of the twin's physical life and also data of the environment
interacting with it.
• Reasoning: drawing conclusions which are consistent with a starting point.
• Learning: using a process to get to the conclusion and provide answers that can
be applied to other domains. Problem solving, finding a solution to the proposed
problem.
• Problem solving. finding a solution to a given problem.

Mortlock et al. (2021) [68] proposed a framework based on graph learning as a

possible way of facilitation of cognition in DTs. This may be used to conduct different
tasks (e.g., to generate new configurations that are necessary in the event of the product
specification change). A graph can show critical relationships and provides a strong
illustration, aiding in deduction or logical solution to problems through visual
representation.

In the first step (Figure 35), the graph formation, products and their properties
are retrieved (e.g., via a query) and organized into a graph, which prepares them for
analysis. The graph operation involves the modelling of intricate mathematical
functions, the aggregation of data, and the creation of condensed representations.
Finally, in the last step, learning objective, the query and the problem to be solved are
defined, as are the metrics and specifications used to optimize and refine the model.

52
 Figure 35: Graph learning framework [68]

The mentioned cognitive abilities can be utilised proficiently in identifying and

handling aberrations in production processes, thus aiding in reducing the adverse
consequences attributed to these anomalies.

Cognitive abilities such as those above can readily be employed to identify and deal
with anomalies in production processes, which in turn can greatly reduce the
deleterious consequences that can accompany such anomalies. To give a brief example,
perception can be used to forecast and recognize anomalies, attention allocated to deal
with them, memory to store relevant information that can be reused, reasoning to
understand their origins and underlying causes, problem-solving to devise efficient
solutions and learning to identify the information that is most important in such
scenarios for use in new instances [66].

Obviously, these capabilities can be utilised with the assistance of supporting tools
(Figure 36):

53
 Figure 36 : Tools supporting cognition [66]

• Memory can be bolstered through the use of a database, a DT capable of

incorporating information from a range of sources, or a knowledge graph, as has
been previously demonstrated. The latter can also foster learning capabilities.
• Perception can be honed through meticulous analysis of data gathered from
various sources.
• Anomalies and control instruments can capture attention.
• Analysis and simulation tools can stimulate reasoning and help evaluate the
impact of anomalies, as well as identify potential solutions, among which lies
the optimal one.

Rozanec et al. (2020) [69] envisioned four components that make a DT actionable
thus aiding in the manufacturing shop floor context: ontology captures information
about entities in the physical world, while a Knowledge Graph enhances the cognitive
capabilities of the DT; data includes detailed information on the elements and
operations of the production process; algorithms, including artificial intelligence
algorithms, enrich the DT with cognitive capabilities and specific behaviours; finally
actions are suggested to users based on advanced analyses performed by the DT.

After having examined the nature of a Cognitive DT and its abilities, we can now

54
explore the synergistic deployment of multiple CDTs in the supply chain context (Figure
37).

Figure 37: Interconnection of CDTs in the supply chain context [67]

In order to establish a connection using Cognitive DTs (CDTs), it is imperative to

create a representation of all assets, processes and operators working in the supply
chain. Afterward, the CDTs must be connected at both intra-factory and inter-factory
levels throughout the supply chain [67].

5.3.6 DTs and sustainability

The sustainability emphasis is growing fast, prompting growing interest very
publicly, along with ever more regulations (e.g., Corporate Sustainability Reporting
Directory [84], requiring companies to report their sustainability progress) and
voluntary certifications (EU Ecolabel, FSC, EPD, etc…). This not simply highlights the
growing interest of companies to both comply with the legal requirements and
demonstrate their sustainability commitment (but also their customers’ increasing
demand for transparency and accountability in this domain, as you can see from Figure
38, showing consumer preferences to buy more sustainable products over the last 5
years worldwide in 2022 [94] .

55
 Figure 38: Degree to which consumers' purchasing behaviour and choices shifted towards buying
more sustainable products over the past five years worldwide in 2022 [94]

For this reason, we will examine how DTs can assist businesses in overseeing
and/or enhancing their level of sustainability.

One of the main questions addressed in this section is:: "How can we apply the just
reviewed capabilities of a DT in a sustainability context?" By conducting research,
Popescu et al. (2022) have explored the impact of each DT attribute on sustainability-
related functions (Figure 39) [70].

Figure 39: DT contribution to sustainability related functions [70]

It is observed that data analytics and visualisation capabilities contribute

significantly to sustainability globally, while cloud processing storage and

56
cybersecurity contribute significantly less. Moreover, the three sustainability-related
functions that are most positively impacted by DT are pollution and waste reduction,
circular economy simulation, and climate change mitigation and control.

Analysing the contribution that DTs can make to achieving the Sustainable
Development Goals (SDGs), it can be seen that DTs can contribute in three ways [74]:

• Efficiency in resource allocation: for example, sensors to detect water leaks

in water distribution systems can be used in a digital replica for predictive
maintenance, thus helping to limit water wastage.
• Safe innovation in green technologies: new clean technologies can be tested
in the virtual world to see if they could cause unintended harm.
• Inclusive partnerships for sustainability: simulations of entire
environments (factories, power grids, farms), accessible to different
stakeholders regardless of their geographical location, facilitate scientific
collaboration and knowledge sharing to address environmental problems.

In addition, by collecting real-time data such as energy and carbon inputs/outputs,

a DT can make Life Cycle Assessment (LCA) much more accurate and faster than current
conventional methods [75].

The second question that arise in this context is, "how a DT could be used to boost
sustainability contribution?" To answer this question, 3 steps were identified, each of
which contributes increasingly to enhancing the sustainability impact of DTs [72]:

• DTs can be employed as a tool to provide information to regulatory bodies

through a sub-model that assesses the impact on the ecosystem. It is evident
that the DT has the capacity to oversee production, hence, it can be integrated
alongside diverse tools such as Life Cycle Assessment, to analyse the worldwide
environmental impacts of production.
• DTs can be utilised to manage assets and form decisions about their applications
throughout their lifecycle, taking into account their impact on the system.
• DTs can serve as a control unit for each particular asset, guaranteeing that the
impact of not only the individual asset but also the entire production system
aligns with sustainability within planetary boundaries.

For these purposes it is essential to ensure interoperability between data, which

57
have to cover information on substances, energy, materials used, as well as emissions
and waste generated throughout the entire life cycle [72] .

A DT which takes into account sustainable production could be used in eco-design,

as well as in the planning and monitoring of manufacturing processes [73].

In eco-design, it is crucial to design products in such a way that most of the

materials used can be recycled. The DT offers a classification model that simplifies
material selection based on their compatibility and degree of recyclability, eliminating
the need for separation. Thus, the DT classifies materials in two ways:

1. The main material and desired level of compatibility are inputted, and
subsequently, the DT selects a range of suitable materials.
2. In the instance where both main and additional materials are utilised, the DT
provides the corresponding degree of compatibility directly.

When planning manufacturing processes, the DT can make decisions about

machine and machining tools, machining parameters and tooling, always considering
the sustainability aspect. For instance, when selecting tooling, the design team can
assess elements that impact the energy required to cut, like tool lifespan, cutting edge
number, cutting time, etc. [73].

Digital twins play also a crucial role in enhancing the sustainability of supply
chains. DTs can contribute to: (1) supply chain visibility by enabling better
monitoring of processes, identifying inefficiencies, and optimizing logistics; (2) carbon
foot print reduction through the optimization of production processes, logistics, and
energy consumption; (3) transparency, allowing exchange of data among supply chain
partners through a shared digital platform [76].

5.4 Enablers, challenges and opportunities

The final section of this chapter has the goal to understand the enablers, challenges
and opportunities that arise from DT implementation (the last phase saw in 5.3).

This section is important because it could help companies to understand the

requirements that can enable the development of the DT frameworks saw in the section
above. We also highlight the fact that the implementation of a DT within a
manufacturing company is not without challenges, but that once these are overcome,

58
several opportunities can arise for the company.

So, here we will illustrate the enablers for DT implementation (5.4.1), the
opportunities deriving from DT adoption (5.4.2) and finally the challenges that a
company can encounter during the implementation phase (5.4.3)

5.4.1 Enablers for DT implementation

Multiple documents in the literature analyse the enabling factors for implementing
a DT and generate an extensive list of requirements. Therefore, as said in 5.1, we
classified them for better understanding:

• The systems and technology category refers to the set of systems and
technologies required to successfully implement a DT.
• The process category pertains to the implementation processes of a DT.
• The people and competences category refers to the requirements that
employee have to meet.

• The culture and strategy category refers to the requirements that the company
as a whole have to meet.

This subdivision is summarised in Table 3.

59
 Enablers

Systems and Simulation [77], [78]

Technologies IoT [77] [79]

Cybersecurity [77]

Big Data processing [77][79][80]

Data storage [79] [80]

Information model [80]

Communication network [79][80]

Data acquisition and cleansing [80]

Time-sensitive data processing [78][80]

Data visualization [78] [81]

VR [78] [79]

Development technologies [79]

Notification system [78]

Process Well-defined implementation plan [77]

Accurate data fillings on enterprise software [77]

Use of physical resources [81]

People and Skills to manage the technologies [77] [81]

Competences Good communication skills [77]

Culture and Management commitment to long-term projects [77]

Strategy Top management support [77]

Capacity to make financial investments [77]

Table 3 : DT enablers

60
Systems and Technology

This first category of enablers comprises of:

- Simulation: testing various scenarios through simulation is essential to make

informed decisions [77], [78]
- IoT: the Internet of Things enables devices to communicate and collect data
through sensors [77][79].
- Cybersecurity: to prevent cyber-attacks and ensure the integrity of data,
cyber security is critical. Protecting sensitive data is always vital for an
organisation [77].
- Big Data processing: big data processing enables the analysis of large
amounts of data, sometimes from multiple sources. It is also a prerequisite for
real-time decision making [77][79][80].
- Data storage: there are two types of data storage that can be used to store the
data collected through the use of sensors and IoT. Relational data storage has
a table structure and is used to handle complex data, while non-relational
databases are used when dealing with less structured data [79][80].
- Information model: the physical object is abstracted using a predefined
information model that represents its specifications of interest. The standard
plays a crucial role in providing the information model to describe different
physical objects in manufacturing [80].
- Communication network: communication between the DT and its physical
counterpart is essential as it is a two-way exchange of information [79][80].
- Data acquisition and cleansing: making decisions based on poor quality data
can lead to wrong conclusions. Since real-world data will never be 100%
accurate, it is important to understand which data should be ignored [80].
- Time-sensitive data processing: it is important to minimize the time gap
between data collection and analysis, especially if the DT is intended to
monitor in real time. Strict latency is essential in such cases [78][80]
- Data visualization: clear visualization of the data collected from the physical
world via dashboards or graphs allows for quicker understanding and
therefore faster decision-making [78][81].

61
 - Virtual Reality (VR): if remote assistance is required, the construction of a
full-scale 3D model of the physical model to be twinned would be a great
advantage [78][79]
- Development technologies: this requirement refers to the technologies
needed to implement a DT [79].
- Notification system: a notification system is required for any type of warning
that the DT needs to send to the operator so that a problem does not go
undetected [78].

Process

This category pertains to the implementation processes of a DT. The enabling

factors within this category include:

- Well defined implementation plan: defining a detailed action plan makes it

possible to stay on schedule, and to analyse every possible scenario. As
mentioned earlier, we must always consider whether or not the
implementation will bring significant benefits [77].
- Accurate data fillings on enterprise software: data in business software
serves as an analytical element for the DT. DTs can also prevent 'incidents' by
analysing historical data. It is therefore important that this data is accurate
[77].
- Use of physical resources: it involves the integration and efficient use of
physical resources within the DT implementation process [81].

People and competences

This section is not about digital technology, but about the people who use it since
they have to meet requirements too. More specifically:

- Skills to manage the technologies: the success of a DT project requires a

thorough understanding of the technologies involved. Operators must be able
to use and manage the key technologies required to effectively develop,
implement and maintain a DT. As this is a relatively new technology, the
company must be open to recruiting new people if it does not already have
people with these skills [77][81].

62
 - Good communication skills: as within any project, more people will be
involved in the implementation. Knowing how to communicate is necessary
for good teamwork [77].

Culture and strategy

In terms of corporate culture and strategy, these also have an impact on the
implementation of a DT. The following requirements should be met:

- Management commitment to long term projects: implementing a DT takes

time and analysis, and the whole team needs to be aware of this when
embarking on the project [77].
- Top management support: the road to implementation is usually long and is
rarely without difficulties. The support of top management will show
confidence in the team dealing with it [77].
- Capacity to make financial investments: few companies have all the
resources needed to implement a DT. Nor is it a cost-free project. Knowing
how to make the right investments will benefit the business significantly [77].

5.4.2 Opportunities deriving from DT utilization

In the context of manufacturing companies, there are several highly significant
benefits of implementing DTs. The strategic implementation of DTs opens up
unconventional vistas in streamlining production processes, managing resources, and
enhancing overall business effectiveness. Given this, in the context of their substantial
role in enabling digital transformation and the progress of manufacturing practices, it
is now important to dwell on the numerous advantages DTs afford. As a result, this
section is devoted to shedding light on the advantages of integrating DTs into
manufacturing firms, and to explain how these technological breakthroughs are
revolutionizing the industrial landscape and creating new opportunities for growth
and improvement.

Firstly, DT accelerates prototyping and redesign processes using simulations

that assay multiple scenarios, shortening design and analysis cycles. It continuously
compares predicted to actual performance, throughout the product lifecycle, leveraging
the DT and the physical twin. It customizes for user needs, using usage data[82].

63
 Secondly, DT DT is cost-effective, generating far less waste because most
prototyping is with virtual resources; this, in turn, reduces aggregate costs. Unlike
traditional prototyping, which uses expensive material and labour, DT also permits
products to be virtually “torn up and tested” without incurring additional material
costs, reducing overall product costs. Simulated testing, with some physical testing,
means less material is “destroyed” [83].

Thirdly, DT's predictive capabilities enhance problem forecasting and system

planning thanks to real-time data flowing between the physical asset and its DT [82].

Furthermore, DT optimizes solutions and improves maintenance, as it projects,

and permits the visualization of defects and wear and tear in manufacturing machinery
or systems, and simulates different scenarios to provide that best solutions (and
maintenance strategies) exist. In this way, it may, for example, take the best possible
decision of when to take a machine out of operation, and also optimize falls over in
service, taking into account the state of the machinery [84].

The accessibility of DT allows remote control and monitoring of physical devices,

overcoming geographical restrictions. This was particularly beneficial during
situations like the COVID-19 pandemic [85].

In hazardous industries like oil and gas or mining, DT's remote access and
predictive nature reduce the risk of accidents [86] .

Lastly, DT contributes to waste reduction by simulating and testing prototypes in

a virtual environment, minimizing material wastage. This virtual probing of prototype
designs under various test scenarios enables the finalization of product designs before
physical manufacturing, aligning with sustainability goals [87].

5.4.3 DT implementation challenges

The available literature leads us to conclude that there are several challenges to
face in the journey to a full implementation of DTs. To explore this strand, we looked at
six papers from the literature that provided insight into the challenges involved in
implementing DTs.

Each of these papers addresses specific challenges related to the implementation

of DTs. Although some of these challenges may overlap, the diversity of perspectives

64
offered allows us to gain a comprehensive overview of the critical issues that may arise
during this technology integration process.

After a careful analysis of the challenges described in the six reference documents,
it was considered important to use a selection criterion that would allow all the
challenges identified by the different authors to be organised more effectively and
clearly. All the criteria analysed, which will be explained in more detail later, and the
categories into which they were divided are summarised in Table 4.

Implementation challenges

Engineering System integration and interoperability [89], [90], [91]

related Necessity for standardization and simplification of processes [89], [90]
challenges
Need of high-performance real-time communication systems [88], [89]

High cost [89], [90], [92], [93]

User interaction [90]

Long-time implementation process [89], [90]

Organizational Multiple stakeholders [89]

challenges
Cultural inertia [89], [92]

Must set realistic expectations, trust and value proposition [89], [93]

Lack of necessary skills and knowledge [90], [93]

Data related Data ownerships [90], [93]

challenges Data variety [90], [91]

Data protection [89], [90], [93]

Data sharing [90]

Table 4 : DT implementation challenges

65
 Engineering related challenges

The lack of system integration and interoperability challenge represents the

inability to integrate different systems or components in a synergistic way, leading to
difficulties in communication and data exchange. Overcoming this challenge means
overcoming differences in operating modes and data formats to ensure smooth
collaboration between the different entities involved in the DT [90]. To overcome this
problem, we could identify open technologies that can be easily integrated with the
production production system [89].

The lack of common standards can hinder interoperability between systems,

effective sharing of data and information, and collaboration between different DT
platforms. Thus, the need for standardisation and simplification of processes
represents a challenge in the implementation of DTs, as it requires the adaptation and
harmonisation of standards and processes [89], [90].

The implementation of DTs requires high-performance real-time

communication systems to enable the immediate exchange of data between the DT
and its physical counterpart. This requires significant resources in terms of technical
expertise and financial investment. In addition, ensuring the security and reliability of
such systems is crucial, as they must handle sensitive data and help control critical
processes in real time [89].

There are also significant costs associated with implementation. Often a

company's resources are not sufficient, and it is often necessary to purchase expensive
sensors, software modules, storage systems and hire new staff to acquire new skills
that were not previously required [89], [90], [92], [93]

Human interactions with machines in the manufacturing environment can be

prone to accidents, safety concerns in the workplace are a significant concern [90].

To avoid all the possible problems listed above, and to find a solution to the various
challenges, long implementation times are often necessary [89], [90].

Organizational challenges

During the implementation of DTs, clear communication about responsibilities,

66
competences, and common objectives is necessary due to the presence of multiple
stakeholders [89].

Cultural inertia can also hinder the implementation of DTs. Workers may be
hesitant to embrace the idea of a 100% reliable digital copy that can account for all
physical world variables, leading to resistance [92]. Therefore, it is advisable to
codesign the DT with operators to ensure complete transparency on the data collected
[89]. A clear and achievable vision must be established, and workers must be trusted
to achieve it [93].

Before deciding to adopt a DT, companies often lack workers with the necessary
skills [92]. This includes Industry 4.0 specialists and digital expertise [91]. One
possible solution is to interact with other companies that have successfully adopted a
DT to fully understand the process, in addition to drawing new resources from outside
[89].

Data related challenges

Regarding data and information flow, data ownership is a significant issue, not
only in relation to DT but also in the broader context of digital transformation [93].
Sharing data across the entire value chain and adding more information over time can
increase intellectual capital. Thus, the issue of data ownership is not insignificant [90].

The data variety needed to fully utilise the capabilities of a DT is not a significant
issue, as data is produced and collected on a daily basis. However, the challenge arises
from the need to integrate, cleanse and fuse the diverse data types [90], [91].

In the digital economy, various types of data are at risk, including personal data,
financial data, information on the development of new technologies, and an
organization's corporate and strategic information. Cyber-attacks are becoming more
frequent and complex, posing a significant threat to data security. The consequences of
such attacks can have a detrimental impact on an organisation's reputation, finances,
and physical assets. Therefore, it is essential to take measures to prevent this risk [90].

Effective information and data sharing is crucial for different actors along the
value chain, both internally and externally. However, corporate policies, cultures, and
people's mindsets regarding data ownership often hinder this process. This presents a

67
significant challenge for DTs, which goes beyond technological and engineering
complexities. Failure to share information can result in the creation of data 'silos' [90].

6 Conclusions
This thesis presented an overview of the role of DT technology in contemporary
manufacturing systems. Classification criteria were used to appraise the broader
literature, and the associated benefits and challenges in its implementation were
examined.

By classifying the documents according to the functions that they serve, the
classification criteria enable to focus on those specific aspects of DTs most relevant to
their area of interest, and in so-doing, provide insight into the various ways DTs can be
exploited to streamline manufacturing activities. These include monitoring and
optimizing production, layout design, sustainability, manage flexibility and their
collaboration with other DTs.

Nonetheless, this study has some limitations. First, the review of literature in this
study is restricted to the Scopus source of articles. The Scopus source itself may not be
enough to cover the entire spectrum of research literature on DTs in the manufacturing
domain. Besides, classification criteria are not exhaustive, and might not cover all the
potential application of DTs. Finally, the provided frameworks in this thesis are not
wide enough to include all possible use cases of DTs in manufacturing.

However, this thesis holds value in the sense that it has explored a variety of
different features of the role of DT in manufacturing systems, and the findings
underline the transformative potential of DTs, and their capacity to effect positive
change in the manufacturing industry, leading to a new age of smart, connected and
efficient production processes.

68
Bibliography
[1] A. C. Pereira and F. Romero, “A review of the meanings and the implications of the
Industry 4.0 concept,” Procedia Manuf, vol. 13, pp. 1206–1214, 2017, doi:
10.1016/j.promfg.2017.09.032.

[2] S. Weyer, M. Schmitt, M. Ohmer, and D. Gorecky, “Towards Industry 4.0 -

Standardization as the crucial challenge for highly modular, multi-vendor
production systems,” IFAC-PapersOnLine, vol. 48, no. 3, pp. 579–584, 2015, doi:
10.1016/j.ifacol.2015.06.143.

[3] M. Javaid, A. Haleem, and R. Suman, “Digital Twin applications toward Industry
4.0: A Review,” Cognitive Robotics, vol. 3, pp. 71–92, 2023, doi:
10.1016/j.cogr.2023.04.003.

[4] Gizem Erboz, “How To Define Industry 4.0: Main Pillars Of Industry 4.0,” Godollo,
Hungary, Nov. 2017.

[5] Statista, “Industry 4.0: in-depth market analysis,” Apr. 2023.

[6] B. R. Barricelli, E. Casiraghi, and D. Fogli, “A Survey on Digital Twin: Definitions,

Characteristics, Applications, and Design Implications,” IEEE Access, vol. 7, pp.
167653–167671, 2019, doi: 10.1109/ACCESS.2019.2953499.

[7] M. Singh, E. Fuenmayor, E. Hinchy, Y. Qiao, N. Murray, and D. Devine, “Digital Twin:
Origin to Future,” Applied System Innovation, vol. 4, no. 2, p. 36, May 2021, doi:
10.3390/asi4020036.

[8] Gartner, “Gartner Top 10 Strategic Technology Trends for 2019.” Accessed: Jan.
24, 2024. [Online]. Available:
https://www.gartner.com/smarterwithgartner/gartner-top-10-strategic-
technology-trends-for-2019

[9] Google Trends, “GoogleTrends - Digital Twin.” Accessed: Jan. 24, 2024. [Online].
Available: https://trends.google.com/trends/explore?date=today%205-
y&q=digital%20twin&hl=it

[10] Global Market Insight, “Digital Twin Market Size By Application (Product Design
& Development, Machine & Equipment Health Monitoring, Process Support &

69
 Service), By End Use (Manufacturing, Healthcare, Automotive, Aerospace &
Defense, Energy, Others) & Forecast, 2023 - 2032,” Apr. 2023. Accessed: Jan. 24,
2024. [Online]. Available: https://www.gminsights.com/industry-
analysis/digital-twin-market

[11] Statista, “Global digital twin market share in 2020, by industry.” Accessed: Jan.
24, 2024. [Online]. Available: https://www-statista-
com/statistics/1296192/global-digital-twin-market-share-by-industry/

[12] Statista, “Global digital twin market size in the year 2020 and 2025, by
industry(in billion U.S. dollars).” Accessed: Jan. 24, 2024. [Online]. Available:
https://www-statista-com/statistics/1296187/global-digital-twin-market-by-
industry/

[13] F. Tao, J. Cheng, Q. Qi, M. Zhang, H. Zhang, and F. Sui, “Digital twin-driven product
design, manufacturing and service with big data,” The International Journal of
Advanced Manufacturing Technology, vol. 94, no. 9–12, pp. 3563–3576, Feb. 2018,
doi: 10.1007/s00170-017-0233-1.

[14] M. Conroy, “Modeling, Simulation, Information Technology and Processing

Roadmap,” 2010. [Online]. Available:
https://www.researchgate.net/publication/280310295

[15] J. Lee, E. Lapira, B. Bagheri, and H. an Kao, “Recent advances and trends in
predictive manufacturing systems in big data environment,” Manuf Lett, vol. 1,
no. 1, pp. 38–41, 2013, doi: 10.1016/j.mfglet.2013.09.005.

[16] A. Banerjee, R. Dalal, S. Mittal, and K. P. Joshi, “Generating Digital Twin Models
using Knowledge Graphs for Industrial Production Lines,” Association for
Computing Machinery (ACM), Jun. 2017, pp. 425–430. doi:
10.1145/3091478.3162383.

[17] E. Negri, L. Fumagalli, and M. Macchi, “A Review of the Roles of Digital Twin in
CPS-based Production Systems,” in Procedia Manufacturing, Elsevier B.V., 2017,
pp. 939–948. doi: 10.1016/j.promfg.2017.07.198.

[18] Hempel, “Healthcare solution testing for future | Digital Twins in healthcare.”

[19] Q. Qi and F. Tao, “Digital Twin and Big Data Towards Smart Manufacturing and

70
 Industry 4.0: 360 Degree Comparison,” IEEE Access, vol. 6, pp. 3585–3593, 2018,
doi: 10.1109/ACCESS.2018.2793265.

[20] “Global Horizons Global Horizons Final Report United States Air Force Global
Science and Technology Vision,” 2013.

[21] J. Guo, N. Zhao, L. Sun, and S. Zhang, “Modular based flexible digital twin for
factory design,” J Ambient Intell Humaniz Comput, vol. 10, no. 3, pp. 1189–1200,
Mar. 2019, doi: 10.1007/s12652-018-0953-6.

[22] K. M. Alam and A. El Saddik, “C2PS: A Digital Twin Architecture Reference Model
for the Cloud-Based Cyber-Physical Systems,” IEEE Access, vol. 5, pp. 2050–2062,
2017, doi: 10.1109/ACCESS.2017.2657006.

[23] M. Mabkhot, A. Al-Ahmari, B. Salah, and H. Alkhalefah, “Requirements of the

Smart Factory System: A Survey and Perspective,” Machines, vol. 6, no. 2, p. 23,
Jun. 2018, doi: 10.3390/machines6020023.

[24] M. Grieves and J. Vickers, “Digital Twin: Mitigating Unpredictable, Undesirable

Emergent Behavior in Complex Systems,” in Transdisciplinary Perspectives on
Complex Systems, 2016, pp. 85–113.

[25] 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, doi:
10.1016/j.ifacol.2018.08.474.

[26] Siemens, “Digital Twin.” Accessed: Feb. 01, 2024. [Online]. Available:
https://www.plm.automation.siemens.com/global/it/our-
story/glossary/digital-twin/24465

[27] F. Tao, Q. Qi, L. Wang, and A. Y. C. Nee, “Digital Twins and Cyber–Physical Systems
toward Smart Manufacturing and Industry 4.0: Correlation and Comparison,”
Engineering, vol. 5, no. 4, pp. 653–661, Aug. 2019, doi:
10.1016/j.eng.2019.01.014.

[28] A. Madni, C. Madni, and S. Lucero, “Leveraging Digital Twin Technology in Model-
Based Systems Engineering,” Systems, vol. 7, no. 1, p. 7, Jan. 2019, doi:
10.3390/systems7010007.

71
[29] W. Yu, P. Patros, B. Young, E. Klinac, and T. G. Walmsley, “Energy digital twin
technology for industrial energy management: Classification, challenges and
future,” Renewable and Sustainable Energy Reviews, vol. 161, p. 112407, Jun.
2022, doi: 10.1016/j.rser.2022.112407.

[30] M. Enders, M. R. Enders, and N. Hoßbach, “Dimensions of Digital Twin

Applications-A Literature Review Dimensions of Digital Twin Applications-A
Literature Review Dimensions of Digital Twin Applications-A Literature Review
Completed Research,” 2019. [Online]. Available:
https://www.researchgate.net/publication/359715537

[31] M. Grieves, “Digital Twin: Manufacturing Excellence through Virtual Factory

Replication,” 2014. [Online]. Available:
https://www.researchgate.net/publication/275211047

[32] R. Stark, S. Kind, and S. Neumeyer, “Innovations in digital modelling for next
generation manufacturing system design,” CIRP Ann Manuf Technol, vol. 66, no. 1,
pp. 169–172, 2017, doi: 10.1016/j.cirp.2017.04.045.

[33] F. Tao, M. Zhang, Y. Liu, and A. Y. C. Nee, “Digital twin driven prognostics and
health management for complex equipment,” CIRP Annals, vol. 67, no. 1, pp. 169–
172, Jan. 2018, doi: 10.1016/j.cirp.2018.04.055.

[34] A. Redelinghuys, A. Basson, and K. Kruger, “A six-layer digital twin architecture

for a manufacturing cell,” in Studies in Computational Intelligence, Springer
Verlag, 2019, pp. 412–423. doi: 10.1007/978-3-030-03003-2_32.

[35] C. Boje, A. Guerriero, S. Kubicki, and Y. Rezgui, “Towards a semantic Construction

Digital Twin: Directions for future research,” Autom Constr, vol. 114, p. 103179,
Jun. 2020, doi: 10.1016/j.autcon.2020.103179.

[36] “Industry 4.0 and the digital twin Manufacturing meets its match A Deloitte
series on Industry 4.0, digital manufacturing enterprises, and digital supply
networks,” 2018.

[37] C. J. Parris, “Minds + Machines: Meet A Digital Twin.” Accessed: Jan. 27, 2024.
[Online]. Available: https://www.youtube.com/watch?v=2dCz3oL2rTw

[38] R. Soderberg, K. Warmefjord, J. S. Carlson, and L. Lindkvist, “Toward a Digital

72
 Twin for real-time geometry assurance in individualized production,” CIRP
Annals, vol. 66, no. 1, pp. 137–140, 2017, doi: 10.1016/j.cirp.2017.04.038.

[39] C. Zhang, G. Zhou, J. He, Z. Li, and W. Cheng, “A data- and knowledge-driven
framework for digital twin manufacturing cell,” Procedia CIRP, vol. 83, pp. 345–
350, 2019, doi: 10.1016/j.procir.2019.04.084.

[40] G. Zhou, C. Zhang, Z. Li, K. Ding, and C. Wang, “Knowledge-driven digital twin
manufacturing cell towards intelligent manufacturing,” Int J Prod Res, vol. 58, no.
4, pp. 1034–1051, Feb. 2020, doi: 10.1080/00207543.2019.1607978.

[41] C. Zhang, G. Zhou, H. Li, and Y. Cao, “Manufacturing Blockchain of Things for the
Configuration of a Data- and Knowledge-Driven Digital Twin Manufacturing Cell,”
IEEE Internet Things J, vol. 7, no. 12, pp. 11884–11894, Dec. 2020, doi:
10.1109/JIOT.2020.3005729.

[42] M. Kumbhar, A. H. C. Ng, and S. Bandaru, “A digital twin based framework for
detection, diagnosis, and improvement of throughput bottlenecks,” J Manuf Syst,
vol. 66, pp. 92–106, Feb. 2023, doi: 10.1016/j.jmsy.2022.11.016.

[43] J. Baalsrud Hauge et al., “Digital twin testbed and practical applications in
production logistics with real-time location data,” International Journal of
Industrial Engineering and Management, vol. 12, no. 2, pp. 129–140, Jun. 2021,
doi: 10.24867/IJIEM-2021-2-282.

[44] Y. Qamsane et al., “A Methodology to Develop and Implement Digital Twin

Solutions for Manufacturing Systems,” IEEE Access, vol. 9, pp. 44247–44265,
2021, doi: 10.1109/ACCESS.2021.3065971.

[45] J. H. Loaiza, R. J. Cloutier, and K. Lippert, “Proposing a Small-Scale Digital Twin

Implementation Framework for Manufacturing from a Systems Perspective,”
Systems, vol. 11, no. 1, p. 41, Jan. 2023, doi: 10.3390/systems11010041.

[46] K. Warmefjord, R. Soderberg, B. Schleich, and H. Wang, “Digital Twin for Variation
Management: A General Framework and Identification of Industrial Challenges
Related to the Implementation,” Applied Sciences, vol. 10, no. 10, p. 3342, May
2020, doi: 10.3390/app10103342.

[47] S. Huang, G. Wang, D. Lei, and Y. Yan, “Toward digital validation for rapid product

73
 development based on digital twin: a framework,” The International Journal of
Advanced Manufacturing Technology, vol. 119, no. 3–4, pp. 2509–2523, Mar.
2022, doi: 10.1007/s00170-021-08475-4.

[48] F. Xiang, Z. Zhang, Y. Zuo, and F. Tao, “Digital Twin Driven Green Material Optimal-
Selection towards Sustainable Manufacturing,” Procedia CIRP, vol. 81, pp. 1290–
1294, 2019, doi: 10.1016/j.procir.2019.04.015.

[49] X. Ma, F. Tao, M. Zhang, T. Wang, and Y. Zuo, “Digital twin enhanced human-
machine interaction in product lifecycle,” Procedia CIRP, vol. 83, pp. 789–793,
2019, doi: 10.1016/j.procir.2019.04.330.

[50] A. Detzner and M. Eigner, “A DIGITAL TWIN FOR ROOT CAUSE ANALYSIS AND
PRODUCT QUALITY MONITORING,” 2018, pp. 1547–1558. doi:
10.21278/idc.2018.0418.

[51] D. Lee et al., “Digital Twin-Based Analysis and Optimization for Design and
Planning of Production Lines,” Machines, vol. 10, no. 12, p. 1147, Dec. 2022, doi:
10.3390/machines10121147.

[52] M. Sommer, J. Stjepandic, S. Stobrawa, and M. von Soden, “Automated generation

of digital twin for a built environment using scan and object detection as input
for production planning,” J Ind Inf Integr, vol. 33, p. 100462, Jun. 2023, doi:
10.1016/j.jii.2023.100462.

[53] H. Guo, Y. Zhu, Y. Zhang, Y. Ren, M. Chen, and R. Zhang, “A digital twin-based layout
optimization method for discrete manufacturing workshop,” The International
Journal of Advanced Manufacturing Technology, vol. 112, no. 5–6, pp. 1307–1318,
Jan. 2021, doi: 10.1007/s00170-020-06568-0.

[54] Necip Akar, “Optimizing Cellular Manufacturing Facility Layout Design through
Digital Twin Simulation: A Case Study,” Industrial Engineering and Innovation
Management, vol. 6, no. 6, 2023, doi: 10.23977/ieim.2023.060601.

[55] Q. Yan, H. Wang, and F. Wu, “Digital twin-enabled dynamic scheduling with
preventive maintenance using a double-layer Q-learning algorithm,” Comput
Oper Res, vol. 144, p. 105823, Aug. 2022, doi: 10.1016/j.cor.2022.105823.

[56] K. Tliba, T. M. L. Diallo, O. Penas, R. Ben Khalifa, N. Ben Yahia, and J.-Y. Choley,

74
 “Digital twin-driven dynamic scheduling of a hybrid flow shop,” J Intell Manuf,
vol. 34, no. 5, pp. 2281–2306, Jun. 2023, doi: 10.1007/s10845-022-01922-3.

[57] R. Ogunsakin, N. Mehandjiev, and C. A. Marin, “Towards adaptive digital twins

architecture,” Comput Ind, vol. 149, p. 103920, Aug. 2023, doi:
10.1016/j.compind.2023.103920.

[58] A. A. Neto, E. Ribeiro da Silva, F. Deschamps, L. A. do Nascimento Junior, and E.

Pinheiro de Lima, “Modeling production disorder: Procedures for digital twins of
flexibility-driven manufacturing systems,” Int J Prod Econ, vol. 260, p. 108846,
Jun. 2023, doi: 10.1016/j.ijpe.2023.108846.

[59] Carlos Henrique dos Santos and Gustavo Teodoro Gabriel, “Decision-making in a
fast fashion company in the Industry 4.0 era: a Digital Twin proposal to support
operational planning,” The International Journal of Advanced Manufacturing
Technology, Jul. 2021.

[60] C. Zhuang, J. Liu, and H. Xiong, “Digital twin-based smart production

management and control framework for the complex product assembly shop-
floor,” The International Journal of Advanced Manufacturing Technology, vol. 96,
no. 1–4, pp. 1149–1163, Apr. 2018, doi: 10.1007/s00170-018-1617-6.

[61] K. T. Park, J. Lee, H.-J. Kim, and S. Do Noh, “Digital twin-based cyber physical
production system architectural framework for personalized production,” The
International Journal of Advanced Manufacturing Technology, vol. 106, no. 5–6,
pp. 1787–1810, Jan. 2020, doi: 10.1007/s00170-019-04653-7.

[62] M. Helu, T. Hedberg, and A. Barnard Feeney, “Reference architecture to integrate

heterogeneous manufacturing systems for the digital thread,” CIRP J Manuf Sci
Technol, vol. 19, pp. 191–195, Nov. 2017, doi: 10.1016/j.cirpj.2017.04.002.

[63] S. Liu, Y. Lu, X. Shen, and J. Bao, “A digital thread-driven distributed collaboration
mechanism between digital twin manufacturing units,” J Manuf Syst, vol. 68, pp.
145–159, Jun. 2023, doi: 10.1016/j.jmsy.2023.02.014.

[64] R. Sahal, S. H. Alsamhi, J. G. Breslin, K. N. Brown, and M. I. Ali, “Digital Twins

Collaboration for Automatic Erratic Operational Data Detection in Industry 4.0,”
Applied Sciences, vol. 11, no. 7, p. 3186, Apr. 2021, doi: 10.3390/app11073186.

75
[65] M. A. Al Faruque, D. Muthirayan, S.-Y. Yu, and P. P. Khargonekar, “Cognitive Digital
Twin for Manufacturing Systems,” in 2021 Design, Automation & Test in Europe
Conference & Exhibition (DATE), IEEE, Feb. 2021, pp. 440–445. doi:
10.23919/DATE51398.2021.9474166.

[66] P. Eirinakis et al., “Cognitive Digital Twins for Resilience in Production: A

Conceptual Framework,” Information, vol. 13, no. 1, p. 33, Jan. 2022, doi:
10.3390/info13010033.

[67] K. Kalaboukas, J. Rozanec, A. Kosmerlj, D. Kiritsis, and G. Arampatzis,

“Implementation of Cognitive Digital Twins in Connected and Agile Supply
Networks—An Operational Model,” Applied Sciences, vol. 11, no. 9, p. 4103, Apr.
2021, doi: 10.3390/app11094103.

[68] T. Mortlock, D. Muthirayan, S.-Y. Yu, P. P. Khargonekar, and M. A. Al Faruque,

“Graph Learning for Cognitive Digital Twins in Manufacturing Systems,” Sep.
2021.

[69] J. M. Rozanec, “Towards Actionable Cognitive Digital Twins for Manufacturing,”

2020.

[70] D. Popescu, M. Dragomir, S. Popescu, and D. Dragomir, “Building Better Digital

Twins for Production Systems by Incorporating Environmental Related
Functions—Literature Analysis and Determining Alternatives,” Applied Sciences,
vol. 12, no. 17, p. 8657, Aug. 2022, doi: 10.3390/app12178657.

[71] H. Hassani, X. Huang, and S. MacFeely, “Enabling Digital Twins to Support the UN
SDGs,” Big Data and Cognitive Computing, vol. 6, no. 4, p. 115, Oct. 2022, doi:
10.3390/bdcc6040115.

[72] R. Miehe, L. Waltersmann, A. Sauer, and T. Bauernhansl, “Sustainable production

and the role of digital twins–Basic reflections and perspectives,” J Adv Manuf
Process, vol. 3, no. 2, Apr. 2021, doi: 10.1002/amp2.10078.

[73] I. Rojek, D. Mikołajewski, and E. Dostatni, “Digital Twins in Product Lifecycle for
Sustainability in Manufacturing and Maintenance,” Applied Sciences, vol. 11, no.
1, p. 31, Dec. 2020, doi: 10.3390/app11010031.

[74] A. Tzachor, S. Sabri, C. E. Richards, A. Rajabifard, and M. Acuto, “Potential and

76
 limitations of digital twins to achieve the Sustainable Development Goals,” Nat
Sustain, vol. 5, no. 10, pp. 822–829, Jul. 2022, doi: 10.1038/s41893-022-00923-
7.

[75] A. Barni, A. Fontana, S. Menato, M. Sorlini, and L. Canetta, “Exploiting the Digital
Twin in the Assessment and Optimization of Sustainability Performances,” in
2018 International Conference on Intelligent Systems (IS), IEEE, Sep. 2018, pp.
706–713. doi: 10.1109/IS.2018.8710554.

[76] S. S. Kamble, A. Gunasekaran, H. Parekh, V. Mani, A. Belhadi, and R. Sharma,

“Digital twin for sustainable manufacturing supply chains: Current trends, future
perspectives, and an implementation framework,” Technol Forecast Soc Change,
vol. 176, p. 121448, Mar. 2022, doi: 10.1016/j.techfore.2021.121448.

[77] A. A. Neto, F. Deschamps, E. R. da Silva, and E. P. de Lima, “Digital twins in

manufacturing: an assessment of drivers, enablers and barriers to
implementation,” Procedia CIRP, vol. 93, pp. 210–215, 2020, doi:
10.1016/j.procir.2020.04.131.

[78] M. Perno and L. Hvam, “Developing a Framework for Scoping Digital Twins in the
Process Manufacturing Industry,” 2020. doi: 10.3233/ATDE200185.

[79] M. Perno, L. Hvam, and A. Haug, “Implementation of digital twins in the process
industry: A systematic literature review of enablers and barriers,” Comput Ind,
vol. 134, p. 103558, Jan. 2022, doi: 10.1016/j.compind.2021.103558.

[80] 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 Integr Manuf, vol. 61, p. 101837, Feb. 2020, doi:
10.1016/j.rcim.2019.101837.

[81] J. H. Loaiza, R. J. Cloutier, and K. Lippert, “Proposing a Small-Scale Digital Twin

Implementation Framework for Manufacturing from a Systems Perspective,”
Systems, vol. 11, no. 1, p. 41, Jan. 2023, doi: 10.3390/systems11010041.

[82] F. Tao, J. Cheng, Q. Qi, M. Zhang, H. Zhang, and F. Sui, “Digital twin-driven product
design, manufacturing and service with big data,” The International Journal of
Advanced Manufacturing Technology, vol. 94, no. 9–12, pp. 3563–3576, Feb. 2018,
doi: 10.1007/s00170-017-0233-1.

77
[83] M. Grieves and J. Vickers, “Digital twin: Mitigating unpredictable,
undesirable emergent behavior in complex systems,” in In Transdisciplinary
Perspectives on Complex Systems, Springer, 2017, pp. 85–113.

[84] E. Glaessgen and D. Stargel, “The Digital Twin Paradigm for Future NASA and U.S.
Air Force Vehicles,” in 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural
Dynamics and Materials Conference<BR>20th AIAA/ASME/AHS Adaptive
Structures Conference<BR>14th AIAA, Reston, Virigina: American Institute
of Aeronautics and Astronautics, Apr. 2012. doi: 10.2514/6.2012-1818.

[85] W. Shen, C. Yang, and L. Gao, “Address business crisis caused by COVID-19 with
collaborative intelligent manufacturing technologies,” IET Collaborative
Intelligent Manufacturing, vol. 2, no. 2, pp. 96–99, Jun. 2020, doi: 10.1049/iet-
cim.2020.0041.

[86] C. Stamford, “Gartner Survey Reveals 47% of Organizations Will Increase

Investments in IoT Despite the Impact of COVID-19,” Gartner. Accessed: Jan. 28,
2024. [Online]. Available: https://www.gartner.com/en/newsroom/press-
releases/2020-10-29-gartner-survey-reveals-47-percent-of-organizations-will-
increase-investments-in-iot-despite-the-impact-of-covid-19-
#:~:text=By%202023%2C%20One%2DThird%20of,survey*%20from%20Gart
ner%2C%20Inc.

[87] M. Singh, E. Fuenmayor, E. Hinchy, Y. Qiao, N. Murray, and D. Devine, “Digital Twin:
Origin to Future,” Applied System Innovation, vol. 4, no. 2, p. 36, May 2021, doi:
10.3390/asi4020036.

[88] I. Ozkaya, “Architectural Concerns of Digital Twins,” IEEE Softw, vol. 39, no. 2, pp.
3–6, Mar. 2022, doi: 10.1109/MS.2021.3130872.

[89] N. Saporiti, V. G. Cannas, R. Pozzi, and T. Rossi, “Challenges and countermeasures

for digital twin implementation in manufacturing plants: A Delphi study,” Int J
Prod Econ, vol. 261, p. 108888, Jul. 2023, doi: 10.1016/j.ijpe.2023.108888.

[90] S. Singh, E. Shehab, N. Higgins, K. Fowler, T. Tomiyama, and C. Fowler, “Challenges

of Digital Twin in High Value Manufacturing,” Oct. 2018. doi: 10.4271/2018-01-
1928.

[91] A. García, A. Bregon, and M. A. Martínez-Prieto, “Towards a connected Digital

78
 Twin Learning Ecosystem in manufacturing: Enablers and challenges,” Comput
Ind Eng, vol. 171, p. 108463, Sep. 2022, doi: 10.1016/j.cie.2022.108463.

[92] M. Gulewicz, “Digital twin technology — awareness, implementation problems

and benefits,” Engineering Management in Production and Services, vol. 14, no. 1,
pp. 63–77, Mar. 2022, doi: 10.2478/emj-2022-0006.

[93] M. Timperi, K. Kokkonen, L. Hannola, and K. Elfvengren, “Impacts of digital twins

on new business creation: insights from manufacturing industry,” Measuring
Business Excellence, vol. 27, no. 3, pp. 433–448, Aug. 2023, doi: 10.1108/MBE-09-
2022-0104.

[94] Simon-Kucher, “Degree to which consumers’ purchasing behavior and choices

shifted towards buying more sustainable products over the past five years
worldwide in 2022,” Statista. Accessed: Jan. 27, 2024. [Online]. Available:
https://www-statista-com/statistics/1377869/global-shift-to-buying-
sustainable-products/

79