This is a repository copy of Digital twin in manufacturing : conceptual framework and case

studies.

White Rose Research Online URL for this paper:

https://eprints.whiterose.ac.uk/183303/

Version: Published Version

Article:
Onaji, I., Tiwari, D., Soulatiantork, P. et al. (2 more authors) (2022) Digital twin in
manufacturing : conceptual framework and case studies. International Journal of Computer
Integrated Manufacturing. ISSN 0951-192X

https://doi.org/10.1080/0951192x.2022.2027014

Reuse
This article is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs
(CC BY-NC-ND) licence. This licence only allows you to download this work and share it with others as long
as you credit the authors, but you can’t change the article in any way or use it commercially. More
information and the full terms of the licence here: https://creativecommons.org/licenses/

Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by
emailing [email protected] including the URL of the record and the reason for the withdrawal request.

[email protected]
https://eprints.whiterose.ac.uk/
 International Journal of Computer Integrated
Manufacturing

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tcim20

Digital twin in manufacturing: conceptual

framework and case studies

Igiri Onaji, Divya Tiwari, Payam Soulatiantork, Boyang Song & Ashutosh
Tiwari

To cite this article: Igiri Onaji, Divya Tiwari, Payam Soulatiantork, Boyang Song & Ashutosh Tiwari
(2022): Digital twin in manufacturing: conceptual framework and case studies, International Journal
of Computer Integrated Manufacturing, DOI: 10.1080/0951192X.2022.2027014

To link to this article: https://doi.org/10.1080/0951192X.2022.2027014

© 2022 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group.

Published online: 24 Jan 2022.

Submit your article to this journal

Article views: 273

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tcim20
INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING
https://doi.org/10.1080/0951192X.2022.2027014

Digital twin in manufacturing: conceptual framework and case studies

Igiri Onaji, Divya Tiwari, Payam Soulatiantork, Boyang Song and Ashutosh Tiwari
Department of Control and Systems Engineering, The University of Sheffield, Sheffield, UK

ABSTRACT ARTICLE HISTORY

The digital twin (DT) concept has a key role in the future of the smart manufacturing industry. This Received 11 June 2020
review paper aims to investigate the development of the digital twin concept, its maturity and its Accepted 5 January 2022
vital role in the fourth industrial revolution. Having identified its potential functionalities for the KEYWORDS
digitalisation of the manufacturing industry, the digital twin concept, its origin and perspectives Digital twin; automation and
from both the academic and industrial sectors are presented. The identified research gaps, trends control; smart
and technical limitations hampering the implementation of digital twins are also discussed. In manufacturing; Industry 4.0;
particular, this review attempts to address the research question on how the digital twin concept cyber-physical production
can support the realisation of an integrated, flexible and collaborative manufacturing environment systems; product lifecycle
which is one of the goals projected by the fourth industrial revolution. To address this, a conceptual management
framework supporting an integrated product-process digital twin for application in digitised
manufacturing is proposed. The application and benefits of the proposed framework are presented
in three case studies.

1. Introduction In manufacturing, the convergence of the virtual

space with the physical operational space, to enable
1.1. Background
the interconnection of virtual elements with their opera-
In recent years, the tremendous growth and innova- tional physical counterparts (cyber-physical integration)
tive breakthroughs in the digital world and the has been a significant challenge for achieving the objec-
increasing integration of information and communi- tives of smart production (Tao and Zhang 2017; Cheng
cation technologies (ICT) with industrial operational et al. 2018). The concept of the digital twin (DT) has been
technologies (OT) has greatly influenced and rede- discussed for over a decade as an approach to tackle this
fined the manufacturing industry. This has enabled problem and in recent has gained much more attention
better energy and resource utilisation, shortened worldwide (Cheng et al. 2018; Zhou et al. 2019). This
time-to-market for products and enhanced manufac- provides the linkage needed to bridge the gap between
turing flexibility (Rosen et al. 2015; Zhang et al. the physical and virtual space in real-time, interconnect
2019a). Innovations like the interconnections of intel- silos of data within a business chain and reinvent the
ligent components within factory floors, Internet of paradigm of demand and supply (Rosen et al. 2015;
Things (IoT), sensor data fusion and cloud computing Schuh and Blum 2016; Yun, Park, and Kim 2017).
(CC) technologies has given birth to a new era of
manufacturing most often called smart manufactur-
1.2. Problem domain
ing/digitised manufacturing (Tao and Zhang 2017;
Yun, Park, and Kim 2017). Several national strategies 1.2.1. Discussion of the problem
have been developed to harness the potentials of Industry 4.0 drives the manufacturing industry into
these emerging technologies/innovations within the a new era of autonomous and intelligent information
manufacturing industries. Examples include Industry exchange, machine control and interoperable produc-
4.0 (I4.0) by Germany, Made in China 2025, Strategy tion systems. One of the key goals of Industry 4.0 is
5.0 by Japan, Advanced manufacturing partnership connectivity and integration of elements within the pro-
and Industrial internet strategies in the United States duction environment (DIN & DKE 2018). This will allow
of America (USA) (Zhang et al. 2019a). companies to build a data footprint through sensors and

CONTACT Igiri Onaji [email protected] Department of Control and Systems Engineering, The University of Sheffield, Portobello St, Sheffield S1
4DW, UK
© 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-
nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built
upon in any way.
2 I. ONAJI ET AL.

monitoring of machines/equipment. However, there are applications, current challenges and lastly, recommen-
challenges associated with the existing structure of man- dations on possible directions for future work. Aivaliotis,
ufacturing systems: centralised control structures and Georgoulias, and Alexopoulos (2019) reviewed the use of
heterogeneity of data due to the variability of manufac- the digital twin in the field of maintenance and health
turing vendor products (Tao et al. 2019). The digital twin prediction. They investigated already existing implemen-
concept is presented as a prospective solution to this tations and proposed ways to improve them. Lu et al.
challenge. This brings us to the research question: ‘How (2020) reviewed digital twin driven smart manufacturing.
does the digital twin concept support the realisation of an In the context of industry 4.0, the development of digital
integrated, flexible and collaborative manufacturing envir- twin technologies, impact, reference model, application
onment as one of the goals projected by the fourth indus- scenarios and research issues of digital twin towards
trial revolution?’. By enabling the integration of both smart manufacturing were discussed. Jones et al.
physical and virtual spaces, a digital twin of (2020) presented a systematic literature review with
a manufacturing system can provide the integrated plat- a thematic analysis on 92 digital twin publications from
form necessary to harness the potential of generated the last 10 years. In characterising the digital twin con-
data. This would see more data-based corrective actions cept, 13 characteristics were presented, namely, physical
taken in real-time to optimise production lines and entity/twin; virtual entity/twin; physical environment;
increase productivity. virtual environment; state; realisation; metrology; twin-
ning; twinning rate; physical-to-virtual connection/twin-
1.2.2. Past work on review ning; virtual-to-physical connection/twinning; physical
Negri, Fumagalli, and Macchi (2017) reviewed the roles processes; and virtual processes.
of the digital twin in the cyber-physical system (CPS)- From the reviews mentioned above, it was observed
based production systems. This paper also presented the that there has been an increasing interest in the digital
history of the concept, the definitions of the digital twin twin concept. However, a variation in the definition,
in scientific literature, its role within Industry 4.0 and description, classification and application of the con-
some recommendations for future research. Kritzinger cept was observed. Despite the disparity in the percep-
et al. (2018) presented a systematic literature review that tion of the digital twin concept by various interest
focussed on the use of digital twin in manufacturing. In groups, the last two years have witnessed many use
the context of production science, they gave a holistic cases in manufacturing (Figure 2) (Tao, Zhang, and Nee
overview highlighting the manufacturing areas in which 2019). This work creates a holistic picture of the
digital twin has been applied, the concepts, enabling research progress made so far within this field and
technologies and level of integration in recorded use identified the following focus areas for future research:
cases. Having criticised the synonymous use of the
terms digital model (DM), digital shadow (DS) and digital (1) Common grounds for varying ideologies to add
twin (DT), they presented these three terms as subcate- clarity to their applicability
gories of the digital twin based on the level of physical- (2) Tracking of the evolution of the concept up to
virtual data integration. Enders and Hoßbach (2019) pre- the current understanding and applications
sented a systematic review providing a comprehensive within the manufacturing sector
cross-industry overview of the digital twin applications. (3) Integrate both product and process digital
Zhang et al. (2019c) published a systematic review that twins to utilize their dependencies in a cyber-
focused on the current state-of-the-art of digital twin- physical production system (CPPS)
ning within the context of product-service systems. It (4) Forging a research roadmap towards achieving
was observed that little work had been done in the the full dividends of the concept within the
context of product-service systems area. Only two stu- manufacturing industry
dies out of the 59 papers focused on product-service
systems. In an attempt to build an understanding of the
1.3. Novelty and contribution to knowledge
development and applications of the digital twin in
industry, Tao et al. (2019) presented a review on digital Past reviews have pointed out challenges and research
twin in industry. The focus was on the key component of foci within this field (subsection 1.2.2) suggesting that
digital twins, current developments, major digital twin there is a need to track the developments made in
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 3

tackling these challenges and find possible applications quality during production. Three case studies have been
of these solutions within the manufacturing field. This included, of which two case studies are based on the
paper makes its contribution by carrying out a thorough proposed framework. Finally, limitations in the applica-
literature review to investigate the potentiality of the tion of the digital twin concept are presented along with
digital twin concept as an integrated platform to pro- proposed solutions.
mote flexibility and integration in manufacturing. In this regard, the rest of the paper is structured as
A flexibility that allows systems to easily adapt to follows: section two presents the research methodol-
changes in product type, quality, quantity and the inte- ogy. Section three presents the discussion on the
gration of the automation information system to support digital twin concept and its application. Section four
data-driven control methods. In this regard, it has cre- presents a digital twin framework for smart manufac-
ated a holistic picture of the research progress made so turing. Section five presents three case studies.
far within this field. This includes tracing the evolution Section six discusses identified technical limitations
and application of the digital twin concept from incep- and proposed solutions. Finally, the study concludes
tion till date, identification of research gaps, trends and with some recommendations for further research.
technological triggers. This work proposed a digital twin
framework as a concept that can integrate both product
2. Methodology
and process digital twins. This product-process digital
twin integration provides the platform for increased This chapter presents the methodology used in the
flexibility, control and management of the production systematic literature review and presents the main
system for easy adaptation to changes in product type/ quantitative findings. Figure 1 shows the methodology

Figure 1. Flow chart of the selection process. Adapted from (Moher et al. 2009)
4 I. ONAJI ET AL.

used to gather the digital twin-related literature. This understanding of the digital twin concept; proposed
method involved a literature search of the Scopus and digital twin frameworks within manufacturing and
ScienceDirect databases, quantitative and qualitative other technologies used to achieve cyber-physical
analyses of the selected papers. The keywords ‘digital integration. It also addresses how the digital twin
twin’ and ‘manufacturing’ with the keyword Boolean concept supports integration, flexibility and colla-
‘AND’ was used for the search. The identification and boration within the manufacturing environment and
collation of the critical studies were done using the lastly, presents a summary of the main outcomes of
steps explained in Figure 1: A total of 168 publications the literature review.
were considered for meta-analysis. Unpublished data
were not considered.
A meta-analysis was done to increase the power and 3.1. The concept of the digital twin
precision of this research outcome and more importantly The digital twin (Figure 3) is a virtual replica of its physical
using statistical measures, a survey of the landscape asset built mainly of structural and behavioural models
enabled the proposition of a map out for future research mainly for basic control, monitoring and evaluation of its
directions (Stapic et al. 2012). This research may be performance (Cai et al. 2017; Martinez et al. 2018).
biased since the choice of which journal article should
be included or not was subjective to the researchers’ 3.1.1. Characteristics of the digital twin
judgment. The definition of quality is dependent on the A digital twin, as shown in Figure 3 & Figure 8 ,
researchers as such the quality of the articles used in the has the following characteristics that differentiate it
study may vary with persons. Figure 2(a) presents the from simulation models (Modoni, Sacco, and Ter
number of publications considered in the highlighted kaj 2016; Cheng et al. 2018; Martinez et al.
years. 2018)
Figure 2 demonstrates the growing research interest
in the area of the digital twin, especially from the year (I) Real-time reflection: highly synchronised with
2017. Figure 2(b&c) shows an increased application of the physical space, the virtual space is a real-
the concept at a systems and system-of-systems (SoS) time reflection of the physical space with a multi-
levels. These findings reveal a growing acceptance of level of fidelity.
the concept as an essential driving force/element for (II) Interaction and convergence: This characteris-
the fourth manufacturing generation. tic is further divided as follows
a. Interaction and convergence in physical
space: It is a complete integration of
3. Digital twin
system phases, elements, services and
This section presents the key findings from the results interaction. Data generated in various
of the literature review conducted by the authors. This phases of physical space are connected
includes the evolution, perspectives and current and accessible.

(a) Depicts trend of (b) Digital twin types (c) Research focus between
digital twin research interest 1994 to 2019

Figure 2. (a) Depicts trend of digital twin research interest. (b) Digital twin types. (c) Research focus between 1994 to 2019.
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 5

Figure 3. General digital twin architecture.

b. Interaction and convergence between his- The definitions of the digital twin given by NASA
torical data and real-time data: Having both and Grieves reflect a broader view of its existing appli-
multi-physics models and data-driven cations (Shafto et al. 2012; Grieves and Vickers 2016).
approaches, a comprehensive digital twin This research sees the digital twin for manufacturing as
contains both domain knowledge and a set of integrated virtual information construct of
timely operational information of the a potential or actual physical system detailed with all
system. necessary minuscule and macro-level of multi-physics,
c. Interaction and convergence between the multi-scale geometric and simulative probabilistic spe-
physical and virtual spaces: The digital twin cifics, suitable for its creation. Virtual models of suitable
is an integrated platform providing granularity with defined functionalities are developed
a smooth bidirectional connection between and integrated into a networked system (Cheng et al.
the two spaces. 2018). The physical system interlinked with this inte-
(III) Self-evolution: The digital twin able to update grated virtual entity updates it with operational data,
its data in real-time automatically, mirrors its thus becoming an exact digital representation of its
physical twin. Parallel connectivity allows com- physical twin (Haag and Anderl 2018). The develop-
parison between the two spaces enabling con- ment of the digital twin for a new physical asset should
tinuous improvement of the virtual models. begin at the design/engineering stage and evolve
through the assets’ lifecycle (Rosen et al. 2015;
Martinez et al. 2018; Qi et al. 2018). For already existing
3.2. The evolution of the digital twin concept
systems, the digital twin can become an additional
Since the public presentation of the digital twin concept, component modelled to reflect the existing function-
authors have argued on the vision of the digital twin alities of the system (Enders and Hoßbach 2019).
resulting in different definitions and applications (Table
1). In recent years, more authors are inclined to the
3.3. Perspectives on the digital twin
notion that the vision refers to a comprehensive virtual
representation with connectivity to the physical and The digital twin concept in an earlier time was applied
functional description of the product/system through- to product design (Zhang et al. 2019b). In recent times
out the lifecycle phases (Yun, Park, and Kim 2017; Cheng it is perceived to encompass the entire business value
et al. 2018; Kritzinger et al. 2018). Table 1 presents some chain resulting in digital twins of products, produc-
identified definitions of the digital twin. It reflects tion process, system performance and services (Leng
a gradual transformation of the concept and its et al. 2019). Despite non-unification in definition and
applicability. description (Table 1), there is a similarity in the key
6 I. ONAJI ET AL.

Table 1. Identified definitions of the digital twin in literature.

No Ref Year Definition of digital twin
1 (Shafto et al. 2012) 2010– “A Digital twin is an integrated multi-physics, multiscale, probabilistic simulation of an as-built vehicle or system
2015 that uses the best available physical models, sensor updates, fleet history, etc., to mirror the life of its
corresponding flying twin”.
2 (Zhang et al. 2019a) 2016 “The digital twin is a virtual representation of the real product. It has the product’s information since the
beginning of the product’s life until the disposal of the product”.
3 (Grieves and Vickers 2017 “A set of virtual information constructs that fully describes a potential or actual physical manufactured product
2016) from the micro atomic level to the macro geometrical level”.
4 (Brenner and Hummel 2017 “A digital copy of a real factory, machine, worker, etc., that is created and can be independently expanded,
2017) automatically updated as well as being globally available in real-time”.
5 (Stark, Kind, and 2017 “A Digital twin is the digital representation of a unique asset (product, machine, service, product service system or
Neumeyer 2017) another intangible asset), that compromises its properties, condition and behaviour using models, information
and data”.
6 (Weber et al., 2017) 2017 “A digital representation of all the states and functions of a physical asset”.
7 (Blum and Schuh 2017) 2017 “A virtual representation of a product on the shop-floor”.
8 (Bohlin et al. 2017) 2017 “A comprehensive physical and functional description of a component, product or system, which includes more or
less all information which could be useful in the current and subsequent lifecycle phases”.
9 (Negri, Fumagalli, and 2017 “The virtual and computerized counterpart of a physical system that can be used to simulate it for various
Macchi 2017) purposes, exploiting a real-time synchronization of the sensed data coming from the field”.
10 (Tao et al. 2018) 2018 “A real mapping of all components in the product life cycle using physical data, virtual data and interaction data
between them”.
11 (Scaglioni and Ferretti 2018 “A near-real-time digital image of a physical object or process that helps optimize business performance”.
2018)
12 (Talkhestani et al. 2018) 2018 “A current, digital model of a product or production system that contains a comprehensive physical and
functional description of a component or system throughout the lifecycle”.
13 (Haag and Anderl 2018) 2018 “A comprehensive digital representation of an individual product. It includes the properties, condition and
behaviour of the real-life object through models and data”.
14 (Liu, Meyendorf, and 2018 “An integrated multi-physics, multiscale, probabilistic simulation of an as-built system enabled by digital threads,
Mrad 2018) that uses the best available models, sensor information, and input data to mirror and predict activities/
performance over the life of its corresponding physical twin”.
15 (Zhuang, Liu, and Xiong 2018 “A virtual, dynamic model in the virtual world that is fully consistent with its corresponding physical entity in the
2018) real world and can simulate its physical counterpart’s characteristics, behaviour, life, and performance in
a timely fashion”.
16 (Sierla et al. 2018) 2018 “Digital twin: a near-real-time digital image of a physical object or process that helps optimize business
performance”.
17 (Kunath and Winkler 2018 ” The Digital twin of a physical object as the sum of all logically related data, i.e. engineering data and operational
2018) data, represented by a semantic data model.”
18 (Tharma, Winter, and 2018 “Digital twin of a real distributed product is a virtual reflection, which can describe the exhaustive physical and
Eigner 2018) functional properties of the product along the whole life cycle and can deliver and receive product
information”.
19 (Eisentrager et al. 2018) 2018 “A digital twin is a digital model of a real object containing lifecycle records and dynamic status data, which are
synchronized in real-time. The model will be used to gain knowledge that can be transferred to the real object”.
20 (Negri et al. 2019) 2019 “An integrated simulation of a complex product/system that, through physical models and sensor updates, ontol
twin”.
21 (Biesinger et al. 2019) 2019 “A digital twin is defined as a realistic model on a current state of the process and behaviour of real objects with its
structure and elements that are connected to it”.
22 (Kabaldin et al. 2019) 2019 “ A set of mathematical models characterizing in real-time the different states of the equipment, the technological
processes, and the business processes in production conditions”.

components of the digital twin: real-time interaction characteristics, behaviours and performance of its phy-
and the replication of physical asset functionalities in sical entities (Cheng et al. 2018). They can be used for
the virtual space etc. simulations during virtual commissioning, monitoring,
For this paper, the digital twin concept can be diagnostics, prediction and control of the state and
applied within the manufacturing industry primarily at behaviour of its twin to support decision-making in the
three levels, namely: product, unit/systems and system development and operation phases of the product, as
of a system (SoS)/shop-floor levels. These virtually repre- well as reflect information continuity throughout the
sent the integration of process, raw material, tools/ product lifecycle (Schleich et al. 2017; Haag and Anderl
equipment, finished product and services, including all 2018). Figure 4 shows increased usage of the concept
static and dynamic compositions of the production sys- within the manufacturing sector.
tem (Qi et al. 2018a). Virtual models used for a digital
twin can include the following physical models (geo- 3.3.1. Digital twin of a product
metric, performance, simulation, etc.), relational rules The digital twin of a product is simply its digital con-
and behaviour models that reflect the state, struct mapping the individual product in the virtual
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 7

space. Its level of functionality and comprehensive- enables the integration of the various lifecycle pro-
ness is dependent on the physical twin and intended cesses of a product, data and resources. This is an
use (Tao et al. 2018; Haag and Anderl 2018). enabling environment for cross-systems and cross-
platform interconnection and interoperability and
3.3.2. Digital twin at a unit/system level optimisation of servitisation (Qi et al. 2018a).
A digital twin at a unit level of the CPS systems is at its Figure 4(a) shows more use of the digital twin to
smallest possible granularity. Such small units include support the unit/systems level. Figure 4(b), demon-
components and equipment (e.g. computer numerical strates the publications trend in the last 10 years.
control (CNC) machines, robots etc.), materials (trans- A large number of publications in 2018 highlight the
port facilities like automated guided vehicles (AGV) and use of digital twin in physical asset’s lifecycle, enabling
other value-added raw materials), and smart environ- an enriched digital twin database built right from the
ments (Kritzinger et al. 2018). The digital twin at design stage to be available all through its lifecycle
a system level is an integrated data-oriented virtual (Jones et al. 2020). Publications in 2019 show an
replica of all necessary process elements. This includes increased discussion of the digital twin in the opera-
all unit-level digital twins of manufacturing equipment, tional phase with a focus on simulation and optimisa-
material flow, operating systems, human resources, tion using real-time operational data. There is
and other value stream elements (Blum and Schuh a congruence amongst authors that digital innova-
2017; Qi et al. 2018a). At the system level of a shop- tions like sensor data fusion, IoT, edge and cloud
floor, models considered include production capability computing technologies, deep learning and machine
models for production capability and characterisation, learning in Artificial intelligence, big data analytics,
process models to link process-related parameters to faster algorithms, increased computational power
product design attributes and mirror the interaction and the availability of more operational data are trig-
between a product and the model of its corresponding gers for the modification of the expectations of the
production process model (Cheng et al. 2018). concept (Lu and Xu 2018; Scaglioni and Ferretti 2018;
Zhang et al. 2019a).
Analysis of the articles identified six key functional-
3.3.3. Digital twin at an SoS level ities inherent in the digital twin applications namely,
The digital twin smart service platforms can be used prognostic and diagnostic analyses, simulation (online
to achieve collaboration between system-level CPSs and offline), control, monitoring/supervision and opti-
and digital twins. This could be a collaboration within misation (Figure 5) (Martinez et al. 2018; Zhuang, Liu,
a factory site where production lines can interact or and Xiong 2018; Zhang et al. 2019a, 2019b). As seen in
different factory sites. Such an integrated platform Figure 5, in recent years, there has been more

(a): Percentage of publications (b): Trend of the digital twin applicability in process/product
based on digital twin type lifecycle phases

Figure 4. (a): Percentage of publications based on digital twin type. (b): Trend of the digital twin applicability in process/product
lifecycle phases.
8 I. ONAJI ET AL.

Figure 5. Digital twin functionalities.

acceptance of the potentiality of the digital twin con- 3.5. Related work on digital twin frameworks in
cept to provide all six functionalities. Figure 5 demon- manufacturing
strates a steady rise in both simulation and
This section reviews the proposed digital twin frame-
optimisation is as a result of the availability of more
works from the literature. Greives’ standard architec-
factual data (operational data) for simulation and
ture for a digital twin model consists of a physical
a better understanding of the physical asset through
asset, virtual replica and connection which is suffi-
mirrored/factual virtual representation/analysis.
cient to establish a cyber-physical interaction
(Grieves and Vickers 2016). This architecture in some
3.4. Current industrial understanding and
use-case has been extended up to a six-dimensional
applications
framework by the inclusion of digital twin data and
The disparity in the understanding of the digital twin services (Tao et al. 2018; Zhang et al. 2019a:2019b).
vision in the industry is seen in its application, as These enable the fusion and evolution of both physi-
shown in Table 2. It is being used along with various cal and virtually generated data and the addition of
phases of the product lifecycle, interconnect business analytical functionalities to the digital twin.
partners and customers. Microsoft Corporation views Tao et al. (2018) presented a digital twin-driven
the digital twin concept as a business transformation product design (DTPD) framework. This serves as
strategy (Microsoft 2017). a guide on the creation of a product digital twin

Table 2. Table of companies’ views and application purpose.

Ref Companies View Purpose
(Schleich et al. 2017) Parametric Technology Real-time connectivity between the Provide efficient after-sales services using digital twin data.
Corporation (PTC) virtual and physical world
(Cheng et al. 2018) General Electric (GE) Modelling of the current state of its To improve the efficiency of performance and health forecast
physical twin of products throughout their lifecycle
(Microsoft 2017). Microsoft Corp. A business transformation strategy Integrate the business supply chain
(Schleich et al. 2017) Dassault Systèmes Improve product design performance
(Schleich et al. 2017) TESLA Inc. To achieve synchronous data flow between vehicles, factory
and other companies.
(Cheng et al. 2018) NASA Interconnection of both physical Used for fault prediction and validation of related systems
assets and an equivalent virtual
replica of the physical system
(Cheng et al. 2018) U.S. Department of An integrated simulation virtual Used for health maintenance of aerospace crafts
Defence replica of the physical asset
(Schleich et al. 2017) Siemens Virtual models of a physical Establish a connection between virtual and physical space to
production system capture the digital process of products from the design
stage to manufacturing to improve efficiency and quality
(Shubenkova et al. 2018) Predix Asset Service based platform for industrial-grade analytics for
Performance operation optimization and performance management
Management (Predix
APM)
(Shubenkova et al. 2018) DXC Technology A software platform used for hybrid car manufacturing
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 9

and the utilisation of its generated knowledge in models is used to enhance the control/management
the product design process. Zhang et al. (2019a) of the physical asset performance and product
proposed a data and knowledge-driven digital twin quality.
framework for a manufacturing cell (DMTC). This
supports an autonomous manufacturing cell using
3.6. Cyber-physical integration using other
data for the perception of manufacturing problems
concepts similar to the digital twin
and knowledge for solving identified problems.
This has five-dimensional space namely the physi- In the last decade, other digital solutions for an
cal, digital, data, knowledge and social space. This integrated manufacturing environment have been
framework is expected to support self-thinking, reported in the literature, for example, the Cyber-
self-decision-making, self-execution and self- physical system (CPS), Digital mock-up unit (DMU),
improving. Cheng et al. (2018) also present the Symbiotic simulation and the Product avatar. They
aims of a smart factory for the fourth manufactur- all attempt to connect the physical space with the
ing generation. In this case, the digital twin con- virtual world. The CPS with sensing, computation,
cept is used to achieve physical connection and control and communication capabilities attempts to
data collection, virtual models and simulations, achieve physical asset integration (Ward et al. 2021).
data and information technology systems integra- The DMU, a system engineering 3D modelling pro-
tion and lastly, databased production operations cess developed based on CAX (CAD/CAE/CAM) tech-
and management methods. These expectations nology uses computer simulations as a replica of
are also embraced by other authors like Ellgass the actual mock-up (Zhang and Li 2013; Ríos et al.
et al. (2018), Qi et al. 2018a) and Zhang et al. 2015). They can imitate the geometrical, physical
(2019a). and behavioural characteristics of the actual pro-
Notably, Stark et al. (2019) applied information duct mock-up thus providing real responses to
factories in the development and operation of exterior prompting as would the actual mock-up.
a digital factory twin. They proposed an eight- Like the DT, real-time interactions with the product
dimensional digital twin model. Four of these enable immersion sensory perception (Ríos et al.
dimensions namely integration breath, connection 2015; Tao et al. 2019(2)).
mode, update frequency and product lifecycle char- The symbiotic simulation system, birthed under
acterise its environment and the other four dimen- discrete event simulation (DES) draws inspiration
sions: CPS intelligence, simulation capabilities, from symbiosis in biology. It emphasises a mutually
digital twin model richness and human interaction beneficial close relationship between a physical sys-
describe its behaviour (capability richness expressed tem and its simulation system (Mitchell & Yilmaz
in levels). Lu et al. (2020) presented a digital twin 2008). Like the DT, the simulation system has access
reference model with three components, namely, an to real-time sensor operational data from the physical
information model, a communication mechanism system. With this, highly accurate simulations of the
and a data processing module. The information physical system are carried out and in turn, the phy-
model has two subtypes: a model for a product sical system benefits from the decisions made based
digital twin and another for a production digital on the outcome of the simulations of several scenar-
twin. ios representing different operational decisions
Existing digital twin frameworks in literature do (Mitchell & Yilmaz 2008). The product avatar concept
not consider the modelling of the interaction with no explicitly significant difference from the DT
between the product and its production processes concept focuses on user-oriented product formation.
(Grieves and Vickers 2016; Tao et al. 2018; Zhang Like the DT, it also considers sensor data and the
et al. 2019a:2019b). As a result, they do not present physical product is digitally represented using several
the integration of the product and process digital models (Ríos et al. 2015). In addition to integration,
twins in this perspective. A CPPS integrated pro- the digital twin offers the advantages of inclusiveness
duct-process digital twin with logically connected of communication, interaction and collaboration
data/control/resources is a suitable digital platform between the virtual and physical space (Tao et al.
where the interaction of the product and process 2019(2)). It serves as a representation of both the
10 I. ONAJI ET AL.

digital and physical properties of the current and the physical world. This mechanism will support the
future state of the product, equipment or process dynamic generation and iterative co-evolution of
(Schroeder et al. 2016). models and big manufacturing data.

3.7.4. Data-based production operations and

3.7. Cyber-physical integration using the digital management methods
twin concept Physical-cyber consistency and synchronisation pre-
Cheng et al. (2018) presented the aims of a smart sent an avenue for more effective utilisation of
factory for the fourth manufacturing generation and generated data for value creation through colla-
analysed them in four contexts namely (i) physical boration. Operational optimisation in factories can
integration and data collection, (ii) digital/virtual be improved through the integration of data-driven
models and simulations, (iii) data and information services and interdependencies of applications
technology systems integration and lastly, (iv) data- allowing on-demand matching and utilisation of
based production operations and management services.
methods. Viewing the capabilities of the digital
twin from these perspectives constructs a more 3.8. Summary of literature review
vivid picture of how it supports integration, flexibility
and collaboration within the smart manufacturing There is a growing acceptance of the digital twin
environment. concept as an essential driving force/element for the
fourth manufacturing generation. There is yet no
common framework for a digital twin model but it
3.7.1. Physical asset integration and data collection
has found applicability in the various stages of the
The creation of ubiquitous interconnections
product/process. The integration of ICT technologies
between physically separated elements/subsystems
and AI into production systems has paved the way for
of a production system based on a collaborative
a lot of possibilities with the digital twin in manufac-
and context-awareness initiative will support data
turing. Six key functionalities inherent in the digital
collection, interaction and interoperation within an
twin applications were identified: prognostic and
integrated environment. Such an integrated opera-
diagnostic analyses, simulation (online and offline),
tional environment is obtainable using the CPPS
control, monitoring/supervision and optimisation.
concept.
The digital twin concept initially was applied to
product design. In recent times, it is perceived to
3.7.2. Digital/virtual models and simulation encompass the entire business value chain resulting
The creation of multi-dimensional models by the inte- in digital twins of products, production process, system
gration of faithful-mirrored virtual models of both the performance and services. The current vision of the
product/processes systematically constructed using digital twin concept in manufacturing supports the
all necessary data (both engineering and operational realisation of an integrated, flexible and collaborative
data) within a closed-loop bidirectional network with manufacturing environment. Supported by a closed-
the physical assets will contain geometric and physi- loop bidirectional communication network, it promotes
cal properties models of elements, response models asset-twin co-evolution through real-time interaction,
of behaviours and logical models of relationships. This control and convergence in three key areas: within the
will enable reliable and synchronous real-time sys- physical space, between the physical and virtual spaces
tems/models co-simulation, correction, modification and between historical and real-time data.
and control.

4. Digital twin framework for smart

3.7.3. Data and information system integration
manufacturing
The virtual models provide the mechanism (relational
rule model) for a methodical integration and unam- This section proposes a digital twin framework based
biguous fusion of cyberspace (all elements/flows/ on the integration of both the product and process
businesses-covered) data with data perceived from digital twins. The methodology for the proposed
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 11

framework is discussed in section 4.2, the actual fra- interaction is presented as the collaboration mechanism
mework is described in section 4.3 and the benefits of between the product/process digital twins and physical
the proposed framework for manufacturing systems assets. This mechanism during production would create
are discussed in section 4.4. new digital twin collaborative functionalities that would
improve both digital twins performances. Its benefits
are further explained in subsection 4.4.
4.1. Background

This paper advocates for the integration of both product 4.2. Methodology for implementing the proposed
and process digital twin to support cyber-physical pro- framework
duction systems (CPPS). Integrating production systems
with ICT technologies and employing artificial intelli- This subsection discusses the integration of the pro-
gence in manufacturing systems transforms traditional duct and process digital twins and also, the approach
production systems into what is now known as cyber- used in utilising the dynamic interaction between the
physical production systems (Negri et al. 2019). The product and its production processes in supporting
CPPS paradigm with more ICT enhancements, scalable flexibility and product customisation.
modular structures and distributed control introduces
autonomous integrated, adaptable production systems 4.2.1. Integrating the product and process digital
that shortens engineering time in production processes twins
and cost of mass customisation. (Qi et al. 2018). The dynamic interaction between the product and its
Lu et al. (2020)’s model talked about the product and production processes is shown in Figure 6 . Modelled
the process information models. This work extends this as a collaboration mechanism, this logical interaction
idea to an integrated product-process digital twin con- can be bidirectional where the process digital twin
cept to harness the benefits in modelling the logical provides data (value addition) for building the pro-
interaction between the product and its production duct digital twin and the product digital twin provides
processes. This interdependence between the product process configurations relative to predefined product
and its production processes modelled as a logical specifications.

Figure 6. Integrated product-process DT showing the interaction between product and process twins.
12 I. ONAJI ET AL.

This mechanism can be implemented using Yrjänäinen et al. 2016; Sierla et al. 2018). To promote
a relational rule model. The integration of the product product customisation and production flexibility using
and process is achieved using the steps below:a the proposed digital twin framework (section 4.3), the
product-centric control method is adopted to enable
● Stage 1: Carry out a logical mapping of the pro- the product twin to influence its production
duct attributes to the respective process services (Eisentrager et al. 2018; Sierla et al. 2018). Product
that generate them. This is done by identifying specifications defined in the virtual space can be linked
the value-added services that build the product to process configurations such that processes/
and its attributes. machines can be configured to handle customer varia-
● Stage 2: Develop the process and product models. tion. This could involve the arrangement of raw mate-
This can be done on the same or separate digital rials, the sequence of processes and resources in
platform(s). Using the same digital platform elim- general (Sierla et al. 2018). Useful time is saved as
inates the challenges/constraints with communi- setup time due to some manpower reorganisation/
cation links between heterogenous platforms. reconfiguration of the production layout is reduced.
● Stage 3: Develop a relational rule model that logi- Figure 7 shows the path (a-b) used by the virtual
cally implements the interaction between the pro- product to trigger the configuration of the physical
cess digital twin services and the product digital system. When the product is selected, its specifications
twin attributes. This establishes the logical con- determine the configuration of the virtual process
nection between the product digital twin and the model which in turn triggers the configuration of the
process digital twin. It maps the product attri- physical system.
butes to the respective service configurations
and manages the logical integration of the pro-
4.3. Proposed digital twin framework for
duct and process data to create an integrated
manufacturing systems
product-process digital twin information model.
This section presents the proposed digital twin frame-
The product twin based on product specifications work that supports the integration of both product
can influence the configuration of the production sys- and process digital twins. The framework proposed
tem through the process digital twin. On the other, the (Figure 8) in this section comprises six components:
production system through the process digital twin (a) Integrated physical assets, (b) Integrated faithful
can provide the product data needed to mirror the product/process virtual models, (c) Intelligent layer,
physical product. With model standardization, hetero- (d) Data layer and (e) Enterprise layer. Like the other
geneity in generated data can be eliminated between frameworks highlighted above, this framework sup-
the digital twins. This would enable the integration of ports the identified six key functionalities inherent in
product and production data/resources. This collabora- the digital twin applications listed in section 3.3. This
tion would improve each other’s performance and in framework attempt to address the following key
general the physical asset. For instance, using the pro- issues: (1) The integration of all interconnected physi-
duct digital twin, the product quality can be monitored cal elements, (2) Ultra-high synchronisation of the
during production hence reducing the production cost virtual space with the physical space that supports
related to quality assurance activities. real-time interactive simulation and (3) Data fusion
covering all elements, flows, business services and
4.2.2. Product-centric Control Method data-driven and application-oriented services integra-
Product-centric control is an emerging agile method tion (Yun, Park, and Kim 2017; Kritzinger et al. 2018).
that is intended to simplify material handling, control, (a)Integrated physical assets: This refers to
product customisation and information usage within a composition of all related manufacturing entities
the supply chain. It uses unique identification for all of the system interconnected, monitored and man-
associated resources which are linked to control aged using advanced automation equipment like
instructions. While the product is in production, it microcontrollers, programmable logic controllers
directly requests material handling and processes (PLCs), human-machine interface (HMI) for human
from service providers within the supply chain (Lyly- inclusion as an asset, computer-aided controls
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 13

Figure 7. An integrated product-process DT showing the path used by the product twin to influence process configuration using the
product-centric control method.

Figure 8. Digital twin framework of a manufacturing system showing closed-loop interaction

and technical processes; information technologies equipment-CNC machines, robots, workforce, trans-
like sensors, radio frequency identification (RFID), ports, smart environments, smart manufacturing
network infrastructures like servers, intelligent rou- devices, sensors and communication gateways
ters, switches, TCP/IP/Ethernet/OPC_UA/MTConnect (Zhang et al. 2019b).
protocols, etc., and software (Qi et al. 2018b). This (b) Faithful virtual models: This involves the creation
enables the automation of repetitive processes of multi-dimensional models by the integration of
that move and exchange information across the faithful-mirrored virtual models. These models sys-
system, enabling process execution, control, data tematically constructed using all necessary data
perception and transmission (Zhang et al. 2019a). (both engineering and operational data) should con-
Asset here could include passive resources like tain models of geometric and physical properties,
work-in-progress (WIP), active resources like models of elements, response models of behaviours
14 I. ONAJI ET AL.

and logical models of relationships (Tao and Zhang (all elements/flows/businesses-covered) data with
2017). These models in a closed-loop network with data perceived from the physical world. This layer
their physical twin will enable reliable and synchro- will support the dynamic generation and iterative co-
nous systems/models co-simulation, correction, mod- evolution of models and big manufacturing data
ification and control. The level of model granularity is (Zhang et al. 2019a). With the interaction of the phy-
defined by the intended functionality and level of sical space with the virtual space, virtual information/
fidelity. Well defined properties of the physical system models can be updated and operated using real-time
can be modelled using engineering principles while data. The integration of physical and virtual data can
dynamic behaviours can be modelled using stochastic be achieved through the co-evolution of models/data
models which are maintained using operational data. using algorithms to generate information instances of
This approach would enable the evolution of the the current status of the physical asset or results of
virtual models to reflect the current state of the phy- simulation analysis (Tao, Zhang, and Nee 2019). This
sical twin, support diagnostic, prognostic and pre- can then be used to update/reconfigure virtual mod-
scriptive analysis. More recent simulation software els or control/influence the operation of the physical
has been developed to support digital twinning with system. Being able to generate and access standar-
the inclusion of OPC_UA interfaces with standardised dised data makes accumulated information accessible
data exchange, better streaming/storage/processing to all connected entities. Special database middle-
of data and communication. For example, Siemens ware providers like Oracle and Microsoft provide plat-
Tecnomatix equipped with 3D visualisation and forms for data processing, security and storage.
OPC_UA connectivity can be used for real-time super- (e)Enterprise layer: Based on the reference model
vision, control and visualisation of the physical twin. (RAMI4.0) of I4.0, the enterprise layer falls under the
(c) Intelligent layer: The physical connection and hierarchy levels. This layer using service systems such
collection of data allow for the creation of ubiquitous as enterprise resource planning (ERP) and customer
interconnections between physically separated ele- relation management (CRM) enables data-based pro-
ments/subsystems of a production system and vir- duction and management methods like service
tually separated models/algorithms. This layer serves enablement, business models and business decision-
as the brain of the digital twin. It integrates all layers making, event and rule handling etc (Cruz Salazar
of the digital twin using a constructive collaborative et al. 2019). Cyber-physical consistency and synchro-
and context-awareness initiative that supports data nisation present an avenue for more effective utilisa-
collection, interaction and interoperation within an tion of generated data for value creation. Operational
integrated environment (Rosen et al. 2015). optimisation in factories can be improved through the
Programs suitable for specific functionalities are incor- integration of data-driven services and interdepen-
porated to perform at a certain level of autonomy. dencies of applications allowing on-demand match-
Even when distributed control and computerisation is ing and utilisation of services (Negri et al. 2019). The
used to improve system speed and reliability, these internet of things (IoT) which allows the connection of
algorithms/models enables every member to become people and things using internet services/networks
aware of the existence of each other and their oper- and edge/cloud computing are potential technolo-
ability. This layer equipped with artificial intelligence, gies that can extend the vertical integration of this
machine learning algorithms and accessible to both layer across the business chain (Brenner and Hummel
enterprise, engineering and operational data can be 2017). It provides service and interface layers needed
seen to generate information needed to support pro- to seamlessly include contributions made by custo-
duction and enterprise decisions (Macchi et al. 2018). mers and suppliers to the activities/decisions made
Thus this presents the digital twin as a supportive within the system (Qi et al. 2018b).
decision-making system with dynamic knowledge
built through a continuous accumulation of its
interaction. 4.4. Benefits of the proposed framework
(d) Data layer: Data and information system inte-
gration is achieved at this layer, enabling a systematic The proposed digital twin framework offers the fol-
integration and unambiguous fusion of cyberspace lowing benefits:
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 15

4.4.1. Integration setup time during the reconfiguration of the produc-

This framework has a generic structure applicable to tion setup to include customer demand variations in
any manufacturing system, is expandable, enables either the same product or when changing products
fast and easy integration of new resources (physical (Sierla et al. 2018). It expedites a new era of produc-
and virtual) and lastly, is robust to uncertainties pecu- tion where last-minute changes to production and
liar to the manufacturing system. Considering the flexibly respond to disruptions and failures caused
industrial automation pyramid from ISA 95, this digital by suppliers.
twin framework attempts to integrate (vertically and
horizontally) all layers by adopting standardised com-
4.4.4. Analytics
munication like the TCP/IP/Ethernet standard, uni-
Data analytics has been a formidable tool used for
form data formats like the JSON format and
evaluating past incidences and the prediction of the
predefined protocols like the OPC_UA and
future state (Schuh and Blum 2016). As shown in
MTConnect protocol (Cruz Salazar et al. 2019). This
Figure 9, the proposed digital twin framework enables
supports the real-time bidirectional connectivity of
the inclusion of both operational and environmental
IoT within the production system, operational data
data for analytics (Rosen et al. 2015; Catapult 2018).
generation, storage and use, bidirectional control
These additional data reflect the real state of the
and online/offline simulation. All six layers of the
manufacturing infrastructure. The inclusion of such
framework are applicable to both product and pro-
information in analyses reduces the impact of pre-
cess digital twins thus, it supports the integration of
sumption or engineering estimations. This would
both product and process digital twins which in
facilitate the enhancement of manufacturing effi-
extension promotes product customisation, process
ciency leading to a more lean and competitive estab-
flexibility and a supportive decision-making system.
lishment (Onaji et al. 2019; Schuh and Blum 2016).

4.4.2. Interconnectivity
The digital twin framework can be used to tackle 4.4.5. Supportive decision-making system
interconnectivity challenges arising due to the lack Built on real-time operational data, the CPPS is capable
of interaction and interoperability between discon- of making its own decisions about its future or support
nected related industrial sites, independent digital external decision-making systems (Zhang et al. 2019a;
models, non-self-controlled applications and isolated Zhou et al. 2019). Mirroring the physical environment
data silos created within the physical workspace and within the digital space enables analyses of the inter-
virtual information space (Blum and Schuh 2017; active dynamism between the product, process and
Cheng et al. 2018). The use of standard data formats environmental factors resulting in a timely response to
and communication specifications like the OPC_UA such changes. Based on analytical outcomes, the digital
and MTConnect protocols promotes uniformity in models connected to the control system can trigger
data. Also, the use of IoT, cloud computing, machine necessary control commands. This is of significant influ-
learning algorithms and internet technologies like the ence on product and process management and optimi-
5 G network to generate, process and distribute rele- sation (Weyer et al. 2016; Jones et al. 2020). The digital
vant information enables communication and colla-
borations along the supply chain (Negri, Fumagalli,
and Macchi 2017).

4.4.3. Production flexibility and product

customisation
In an integrated product-process digital twin plat-
form, the virtual product influences its production
as product specifications defined in the virtual space
influences the configuration of equipment/produc-
tion layout. When product customisation is auto- Figure 9. Analytics in an interactive digital twin. From: (Catapult
mated, it reduces human intervention and reduces 2018).
16 I. ONAJI ET AL.

twin integrated platform enriched with a standardised Table 3. Festo CP smart factory stations and the task each
semantic description can retain relevant data through- performs
Station Task
out its lifecycle (Schuh and Blum 2016; Cheng et al.
Top case (Station 1) Place the top cover of the phone on the carrier
2018). Measuring (Station Inspects the workpiece on the carrier
4)
Bottom case (Station Loads the carrier with the bottom cover of the
5) phone
5. Case studies Press (Station 6) Couples the top and bottom cover by pressing
them together
Heat tunnel (Station Heat workpiece to a predefined temperature
This subsection presents three case studies. The 7)
authors were involved in the development of the Output station Removes finished product from the production line
(Station 8)
first two case studies based on the proposed digital Bridges (Station 2&3) Transfer terminal the workstations
twin framework in Section 4. These case studies pre- Robotino Transfer product between workstations
sent applicative evidence as a contribution to reinfor-
cing the diversified use of the proposed digital twin
framework/ideas to encourage more investment and
further adoption in existing industrial infrastructures. (PLC). The system is managed using the manufac-
The first case study is based on the Festo cyber- turing execution system (MES) linked to a database
physical smart factory, the second is based on (ACSE 2018). It is composed of two production
a Pharmaceutical continuous crystallisation system Islands and a logistics network of conveyors, carriers
and the third is based on a Virtual X-Ray of electric for conveying the flow entity and an autonomous
motors. The first and second are based on facilities vehicle (Robotino). The workstations are built with
within the University of Sheffield and reflect systems/ six stations and two transfer bridges (Table 3).
structures within the manufacturing industry. Further details about the Festo CP smart factory
can be found in (ACSE 2018).
Production orders are made from the MES.
5.1. Discrete-time CPPS based on the Festo A workpiece is introduced at the ‘Top case’ station.
cyber-physical (CP) smart factory This is inspected at the ‘Measure’ station and then
transported through the rest of the stations: ‘Bottom
5.1.1. The Festo cyber-physical smart factory
case’, ‘Press’, ‘Heat tunnel’, ‘Bridges’ and ‘Output’ sta-
Figure 10a is the integrated physical asset. It is an
tions in that order. At the ‘Output’ station the fin-
Industry 4.0 compact CPPS used for teaching and
ished/rejected product is removed from the
the development of smart industrial automation-
production line. Table 3 gives a list of the Festo CP
based skills. Its resources have been interconnected
smart factory stations and the task each performs.
and monitored over a TCP/IP/Ethernet/OPC_UA net-
The project objective is to build a supervisory digi-
work infrastructure using sensors and RFID technol-
tal twin that supports process monitoring and experi-
ogies. It adopted a distributed control with each
mental analysis. This model is fed with live streams of
module having its Programmable logic controller

Figure 10. (a): Festo CP smart factory. (b): DES digital twin of the Festo CP smart factory.
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 17

data from the sensory devices on the physical asset. of the virtual model, input, output/display ele-
This data can be used to investigate bottlenecks in the ments for data. These graphic user interfaces
processes and their impact on the final product. (GUI) elements control the virtual simulation
and online connection to the physical system.
5.1.2. Methodology (d) The data layer development: This involves the
The methodology used in developing the digital twin of inclusion of data storage elements like tables,
the Festo CP smart factory involved the following steps: global, local variables and interfaces for visua-
conceptual model design, virtual model development, lising information. Both operational and virtual
control panel development, data layer development, data were stored for analytical use.
intelligent layer development and lastly, verification (e) Intelligent layer development: This involves the
and validation process. This approach allows the gradual inclusion of algorithms to generate virtual data,
build-in of the complexity of the existing physical process data to provide required information
system. during controlled experiments. This extracted
information can be used to manage or improve
(a) Conceptual model design: This involved the defi- the operation of the physical system.
nition of the project objective, system definition (f) Model verification and validation: These steps are
and the construction of a conceptual model. carried out intermittently throughout the devel-
This entails the identification of the processes opment stages. All logic and modelled operations
and product attributes to be digitised, physical are debugged and tested to ensure they are
data, process behaviour/functionalities to be error-free and built following the project design.
modelled/visualised and lastly, investigate the
accessibility of the existing system architecture 5.1.3. Developed DES digital twin model of the
to identify how it can be digitally assessed. physical processes
(b) Virtual models development: This includes the pro- The developed DES digital twin model of the Festo
cess and product simulation models whose gran- CP smart factory (Figure 9)), inherits predefined
ularities were defined by the functionalities and behaviours and interaction of the system with its
modular structure identified in the first step. This control algorithms developed to manage the mate-
involved building the 3D models, implementing rial flow. This was achieved using the predefined
the relational rule model that established the material flow objects (Table 4) of the simulation soft-
interaction between the product attributes and ware. Figure 11(a) shows the as-built digital twin
process twins services, control codes and simula- architecture, Figure 11(b&c) shows the DES digital
tion flow. The Siemens NX was used to build a 3D twin of the Festo CP smart factory showing pro-
CAD model (Figure 10), and the Tecnomatix plant cess flow.
simulation (a platform for agent-based/ discrete
event simulations(DES)) was used to construct 5.1.4. Developed DES product digital twin model
the DES model of the processes (Figure 11 The product is a simple composition of a top and
(b&c)). The software used here were selected bottom case for a phone. The process flow analysis
because it got impressive visualisation, the provided details on the services and their weight on
OPC_UA protocol and most importantly can be the attributes of the product. This information is
used for analytics to achieve logistic process
improvement, material flow optimisation and
Table 4. Main objects of the Festo digital twin model.
efficient resource usage (Onaji et al. 2019). This
Material
stage also took care of the creation of the S/no flow objects Number Justification
OPC_UA interfaces used to establish the bidirec- 1 Stations 8 Process execution
2 Footpath 4 Robotino path
tional communication between the physical sys- 3 Conveyors 16 Transport-conveyors
tem and virtual platform. 4 Operator 1 Robotino
5 Workspace 2 Robotino docking stations
(c) Control panel development: This stage involves 6 Checkbox 12 Position sensors at stations and bridges
the addition of control elements like push but- 7 Source 1 Part creation and intro into the system
8 Store 1 Product from the system is stored for removal
tons linked to control instructions/algorithms
18 I. ONAJI ET AL.

(a) As-built digital twin architecture (b) Top-view of the DES digital twin model of the Festo CP smart
factory showing process flow

(c) 3D view of the process DT model

Figure 11. Overview of the developed digital twin of the Festo CP smart factory.

Figure 12. Product digital twin visuals.

conveyed along the production line using RFID tech- 5.1.5. Results and discussion
nology mounted on the carriers. The RFID chip on the Experiments carried out revealed more details on iden-
carrier enables a progressive transmission of the pro- tified bottlenecks and enabled corrective maintenance
duct data as the physical product is processed. This and strategy proposition to improve production.
enables a real-time synchronised update of the pro- Identified bottlenecks included increased friction
duct digital twin. Figure 12 presents the product digital between carriers and conveyors, delay in unloading
twin visuals: a geometric model and extracts of its finished products from the ‘Output’ station, waste of
information model. process time on rejected workpieces and lost process
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 19

(a): The impact of travel time on process time (b): The impact of travel time on total throughput

(c): Impact of robot travel time on total throughput time

. (d): Impact of robot travel time on total throughput per hour
.

(e): The impact of the output station storage

capacity on throughput.

Figure 13. Experimental results revealing more details on identified bottlenecks and system behaviour.

time in heating the ‘Heat tunnel’ station. Figure 13(a&b) 5.1.6. Benefits of the Festo smart factory digital twin
presents the results on the impact on process time and The digital twin developed has expanded the sys-
total throughput by the identified station with increased tem’s research/experimental capabilities for diag-
travel time due to friction on its conveyor. Figure 13(c&d) nostic/predictive analyses, investigate business
presents results on the impact of using the robot at cases and the impact of modification/upgrade
varying travel-time on production total throughput decisions, determine optimal production config-
time. Figure 13(e) presents results on the impact of the urations and monitoring. It has also introduced
output station storage capacity on throughput. Lastly, some level of flexibility in teaching/training. The
offline simulations were carried out to determine the constraint of space and accessibility by a large
impact of controllable system parameters on produc- number of students to the physical system has
tion. Table 13(f) presents the impact of robot travel time been tackled using the DES digital twin which
on the total throughput time and throughput-per-hour. can be used offline/online. Current industrial
These results supported the strategy proposed to expectations like the Industry 4.0 concepts and
keep its store available during production to improve technologies can be taught safely especially in
the system throughput and identified optimal system covid-19 type situations where physical distancing
configurations for production. is vital to safety.
20 I. ONAJI ET AL.

5.1.7. Future work 5.2.2. Methodology

The next phase of the project would upgrade the Using the same methodology in case study one, the
current digital twin to an interactive and immersive digital twin for this system was built using the
predictive digital twin, which reflects the complexities PharmaMV software (Figure 14(b))). The project objec-
and uncertainties experienced in real manufacturing tive was to develop a digital twin that supports pro-
environments and possesses decision-making cap- cess monitoring and control, data visualisation,
abilities. The plan includes an extension of their func- optimisation and multivariant analysis. Its digital
tionalities to real-time control from the virtual space, twin serves as an integrating platform for the soft
including a more robust data layer to implement real- sensor, model development and control design. The
virtual data fusion and product-centric control. PharmMV software suite was selected because it
meets the regulatory requirements of the pharmaceu-
5.2. Continuous-time production system tical industry. It combines both multivariable monitor-
ing techniques that can be used for root cause
5.2.1. The pharmaceutical continuous crystallisation analysis and model predictive control functionality
system for maintenance/improvement of operational effi-
)Figure 14(a) is a highly reliable laboratory environ- ciency/product quality. Table 5 presents the identified
ment used for the nucleation and growth of crystals technical requirements needed in the development of
with consistent properties. It consists mainly of the digital twin and what techniques were used to
a dissolution tank, a chiller and temperature control achieve them.
unit, and a product tank.
Chemical solutions stored in tanks are mixed by 5.2.3. Experimental results
passing them through temperature-controlled The pharmaceutical continuous crystallisation system
jacketed tubes with baffles. Oscillation units within digital twin was used in testing the Model predictive
the tubes are used to mix the solution while cooling controller (MPC) for particle size distribution during
slowly to form crystal as the output of the whole the crystallisation process. This was a test to verify the
process. Figure 15 presents a block diagram of the controller design for a bigger system integrated with
continuous tubular crystalliser showing manipulated the same crystallisation system. Figure 16 presents
variables (MVs) and controlled variables (CVs). the test results. Based on sample data, the controller’s
The main limitations with the continuous tubular active prediction (dark green) was able to trace the
approach of crystallisation are the difficulty in (1) setpoint (red) to yield a stable seed concentration
Control due to supersaturation, temperature, mixture, (light green). It also presents the process analytical
sampling, (2) Implementing process analysis technol- tools (PAT) connected to the soft sensors used to
ogy (PAT) and (3) Blockages in the tubes due to foul- measure other quality attributes that cannot be mea-
ing and sedimentation (Zhang et al. 2017). sured by the physical sensors.

(a): Crystallisation system (b): DT of the crystallisation system

Figure 14. (a): Crystallisation system. (b): DT of the crystallisation system.

 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 21

Figure 15. Schematic of a Continuous tubular crystalliser showing its MVs and CVs.

5.2.4. Benefits of the pharmaceutical continuous 5.3. Virtual X-ray of electric motors
crystallisation system digital twin
5.3.1. Digital twin of an electric motor
With the digital twin, the following functionalities
Developed by Siemens, a virtual X-ray of electric
became accessible to users of the facility: setpoint
motors is a digital twin that enables monitoring of
entry for temperature sensors in the crystalliser
the real-time performance of an electric motor by
tubes, visualisation of crystalliser results like tempera-
utilising thermal simulations to obtain information
ture trend, control performances and concentration;
about temperature distribution inside a motor
crystal size control; availability of data; setting of par-
(Bernard and Sandra 2018). Figure 17 shows a table-
ticle size distribution; setpoint entry for solution con-
sized demonstrator of an electric motor and its digital
centration value; management of the MPCs controller.
twin. Researchers at Siemens Corporate Technology
From a business perspective, the digital platform as
(CT), developed this digital twin with virtual sensors to
a test and training environment supports experimen-
measure and monitor the temperature of the motor
tal analysis without obstructing daily business activ-
components during operation. Using an augmented-
ities. Also, it reduces the financial burden due to the
reality headset to view the demonstrator of the motor
cost incurred from resources used in physical experi-
enabled the user to view the simulation of the motor
ments and training.
and its interior with a real demonstrator superim-
To save cost and also increase student accessibility
posed over it. Colour codes indicated the temperature
in covid-19 like situations, students now have virtual
levels. Detailed information about the temperature
training/experimentations in product/process devel-
distribution inside the motor can enable the operator
opment, characterisation, and reverse engineering of
to decide when the motor is cool enough to be
new products of high-value manufacturing. For exam-
switched on again. This information can help prevent
ple, a methodological examination of design/modifi-
unnecessary downtimes and hence can dramatically
cation decisions, investigation of constituent
lower operating costs.
elements/factor and their impact on outcomes. Also,
several control techniques can be tested to ascertain,
5.3.2. Methodology
which is best to achieve the desired controllability
Mathematical models captured the geometry and
and productivity, support real-time simulation and
material characteristics of drive units to create
emulation of crystallisation processes and products.
a digital twin of each component. Researchers at
Siemens used mathematical reduction processes to
derive abstract models. These models could be calcu-
lated much faster and with fewer deviations in preci-
Table 5. Technical requirements identified and techniques used
in building the digital twin. sion as compared to traditional simulation tools.
S/ Simulation models were then developed with virtual
No Technical requirements Actualisation techniques sensors inbuilt in the motor. These virtual sensors
1 Crystal size distribution System identification
2 Crystal production yield PID Control generated data that was compared with the data
3 Temperature of the mixing tank Model predictive control from sensors on non-moving components.
4 Temperature gradient of crystalliser Experimental data
tubes
5 Flow rate at the incoming of the Identify relevant variable 5.3.3. Benefits
crystalliser
6 Implement PID control Implement a Recursive least This work is useful for large electric drives where the
square (RLS) model
7 Experiment with MPC State-space controller
temperature inside the motor in operation can reach
8 PRBS system identification Digital twin mock-up up to 1000 degrees Celsius, creating a risk for
22 I. ONAJI ET AL.

Figure 16. MPC Controller testing using the simulator environment and Process analytical tool (PAT) connected to soft sensors used
for measuring key attributes not accessible to physical sensors.

deformation of material inside. The manufacturing 6. Technical limitations and solutions

process of electric machines is reliant on strict toler-
The research identified some technical limitations
ances that have decisive influences on the operating
hampering the implementation of the current vision
behaviour of the machine. Therefore, to improve or
of digital twin in achieving closed-loop synchronisa-
better control the manufacturing processes for elec-
tion between the digital and physical space/cyber-
trical machines, few attempts have been made
physical fusion (Qi et al. 2018; Tao et al. 2019). These
towards the development of digital twin in EM man-
technical limitations and proposed solutions are
ufacture. One example is current research by the
described in detail below:
Advanced Manufacturing Research Centre (AMRC) in
Sheffield towards the development of a digital twin of
(1) Lack of quantifiable metrics of uncertainty in digi-
an automated winding process for an electric motor
tal twin models, and unresolved uncertainties in
(FEMM Hub Annual Report 2021). Another work by
the prediction of complex systems (Schuh and
Weigelt et al. (2019), which has demonstrated the
Blum 2016; Jones et al. 2020): Uncertainties are
digital twin of the linear winding process based on
unique to systems due to the variability in the
the explicit finite element method, for the optimiza-
conditions that creates them. No two products
tion of rolling processes.
from the same production line are identical in

(a) Stepper motor (b) Digital twin

Figure 17. Virtual X-ray of electric motors. From: (Bernard and Sandra 2018)
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 23

performance. The use of data-driven models is (1) Variance in the framework of the digital tools
presented here as a solution to this challenge. used to achieve virtual confidence, i.e. linking
The availability of microchips, digital tags and models with data from the factory floor (machine
sensor technologies like the RFID technology or sensor data): One major challenge with the
has expanded data generation beyond geome- digital tool used to achieve virtual confidence
trical measurements (Zhou et al. 2019). This has lies in the variance of the functional/data
made scanning easier and quicker with reliable semantics/communication frameworks (Tao,
tracking capabilities to communicate operational Zhang, and Nee 2019). Interaction/access to
information (Brenner and Hummel 2017). heterogeneous digital platforms/data sources
Integrated product-process models can then be is limited because they are dependent on the
fed with real-time data relating to their opera- integration endpoints and the capabilities of
tional status such as environmental conditions, the database they are built to work on. This
system performance, product quality etc. (Zhang has limited the integration of tools from differ-
et al. 2019c). Such real-time instance data can ent vendors needed in implementing cyber-
help in identifying the pattern of change in the physical integration that supports real-time
behaviour/characteristics of the system/product. communication for smart production (Blum
The use of machine learning algorithms can be and Schuh 2017; Zhang, Tao and Zhang 2017;
used to engineer unique patterns/metrics that Cheng et al. 2018).
can be used to maintain the health of the system.
(2) Virtual confidence: This poses to be a challenge A potential solution is for middleware vendors to
due to the multi-complexity of manufacturing encourage software collaborations by adopting
systems resulting in engineering estimation and widely accepted communication protocols. An exam-
presumptions. Data from the factory floor is com- ple includes the collaboration between MTConnect
pared with data from the virtual model to inves- and OPC_UA communities to provide an MTConnect-
tigate the behaviour of a machine in operation OPC_UA specification that improves the interoper-
against expected behaviour. The use of real-time ability and consistency between both standards. This
data instances in the virtual platform increase capacity is transferable to all manufacturing technol-
virtual confidence (Grieves and Vickers 2016). ogies, equipment, devices or software implementing
these standards. The integrated process-product digi-
Engineering models are known to be a very effec- tal twin as an aggregate model requires more colla-
tive representation of well-known processes. Data- boration between machines/equipment and virtual
driven models are stochastic and tend to represent entities. Communication constraints associated with
the variability of a system. This would be an effective using heterogenous digital platforms in building the
means of representing the product-process interac- models can be handled by either using the same
tion and the progressive development of the product. digital platform or using platforms with unified data/
The management of virtual models and connectivity communication semantics that supports real-time
to the physical twin would be improved if they are communication like the OPC-UA/MTConnect.
built to the lowest possible level of granularity with Another potential solution is the standardisation of
unique identification, functionalities and control. The information model semantics. The systematic
combination of both model approaches would approach towards the development of semantics for
increase the functionality of the integrated digital information models would encourage a continuous
twin. effort towards the conformance and usage of the
More research should be done to establish the same standards (Zhang, Xu et al. 2019). This would
benefits of using various virtual model approaches allow modelling platforms to effectively represent
either separately or combined in the digital twin. It both the product and process composition. For exam-
is interesting to know what levels of engineering and ple, the ISO 10303 standard provides a neutral data
data models combination can be achieved. Trade-off structure enabling CAD systems to exchange product
analysis in use cases would highlight the advantages/ data. ISO 14649 and ISO 10303–238 standards uses
limitations this combination presents. modern associative language enabling direct
24 I. ONAJI ET AL.

connection between CAD design data for machining links are needed to ensure such interactions are seam-
and downstream fabrication processes (Lu, Xu, and less and aligned to the system operation. Also,
Wang 2020). designed algorithms should adopt object-oriented
Software vendors in collaboration with researchers structures in ways that allow human inputs as part
should expand the capabilities of their products. of their operational blocks. State-of-the-art technolo-
Newer versions of existing digital twinning software gies for human-machine interface includes augmen-
with collaborative capabilities that supports the com- ted and virtual reality, natural voice processing and
bined representation of process flow and product gesture control. The concept of immersion stands to
behaviour is needed. The idea of a closed-loop inte- be an effective approach for virtual human-machine
grated digital twin that recreates the product-process interaction.
interaction using enhanced 3D visualisation and data Voice interaction is the most effective and quickest
would be an added advantage for more investigative means of expression for human beings (Nagabushanam,
results in diagnostic analysis. George, and Radha 2020). Nagabushanam, George,
and Radha (2020), highlights advancements in
(1) Lack of an explicitly defined ontology: A closed- Natural language processing (NLP) using neural net-
loop supply chain network would need an worked-based methods. This has been applied in
explicitly defined ontology. Providing an overt deep learning (DL) and long short-term memory
formal specification of the network conceptua- (LSTM) algorithms with a certain level of accuracy
lisation and standards enables seamless busi- and efficiency. In recent times, advancement in
ness integration between trading partners (Lu remote sensing technologies has contributed to better
et al. 2020). Ontologies have an essential role to gesture recognition. Technologies like ambient light,
play to ensure adequate flow of information, cameras and image processing, sound and wearable
reuse of data between project phases, easy devices, Radiofrequency (RF) and mmWave radar are
information accessibility, integration of process been used to capture operator activities. With faster
and product models with enterprise resource communication, sensing and processing capabilities,
planning (ERP) systems, data communication the development of gesture training and control is
through the network and reading data stored a promising development for the digital twin system.
in electronic tags and databases (Cai et al. 2017;
Negri et al. 2019). Blum and Schuh (2017) dis- (1) Lack of professional skills sets: The fourth indus-
cussed several ontologies for IT systems that trial revolution comes with a wave of new tech-
use RFID technology to achieve the smart link- nologies demanding new skill sets to manage
age between the virtual and physical world. them. To create more ways for the actualisation
(2) Challenges in the inclusion of human functional- of the digital twin concept, the manufacturing
ity in the virtual space: Cyber-physical integra- industry continuously needs to liaise with aca-
tion in manufacturing also involves the demia to make fast advancements towards
inclusion of human functionality resulting in these emerging technologies, contextual and
more human-machine interactions. The old social resources (Ward et al. 2021). The industry
control methods involve more human control provides the funding, expertise, application
which limits the autonomy of the system and knowledge and related field data. Academia
restrict human-machine collaborations. provides the technological know-how like
Another challenge here involves the difficulty advanced methodologies from mathematics,
in transferring human operations into machine control and computer engineering and takes
procedures to be handled by the machine to responsibility for grooming the new generation
increase precision and performance. workforce equipped with interdisciplinary skills.
These will result in a progressive transformation
Digital twinning promotes human-machine colla- of their industrial/research environment, and
borations where precision is managed by the work method. A collaboration between these
machine and certain decisions are handled by the two sectors creates a versatile learning environ-
operator. More effective and fast communication ment with a bidirectional channel for
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 25

knowledge sharing and transfer resulting in the evolution of digital twin data. The application, metho-
combination of theoretical knowledge with dology and benefits of the proposed framework were
industrial practices. illustrated in two case studies. The case studies demon-
(2) Challenges in managing big data, defining seman- strated that within manufacturing, the digital twin
tic data models, data management systems and serves as a medium for achieving cyber-physical integra-
scalable databases for data storage on a single tion through bidirectional interaction, data analytics and
platform, the integration of existing simulation linking of information silos all through the product cycle.
packages and semantic interoperability of data The proposed framework allows automated configura-
from heterogeneous sources: Data models mostly tion of production setups using the virtual product spe-
used in engineering and simulation tools are not cifications. Finally, technical limitations and proposed
compatible, resulting in data silos within produc- solutions for the implementation of the digital twin
tion systems (Macchi et al. 2018). There is the issue concept in manufacturing were discussed.
of insufficient details to meet all parameter needs. The digital twin concept is still evolving. More
Also, the proprietary format limits the amalgama- research on digital twin modelling framework, digital
tion of relevant engineering data and models for twin modelling tools and applicative demonstrations
each object. The heterogeneity of gathered data from different fields using integrated product-process
from both physical and virtual spheres poses to be digital twin would prove its diversified usefulness,
a challenge. The proposed DT framework sup- highlight more business benefits, encourage more
ports an extensible framework for data acquisition investment and further adoption in existing industrial
(data gathering, storage, organisation and distri- infrastructures. The idea of controlling physical assets
bution) and analyses for all parts of the system. Lu from virtual models presents more research
and Xu (2018) presents semantic web technology, opportunities.
an evolution of the World Wide Web technology
as a distributed and scalable standardised inter-
face for the surrounding systems. This creates Acknowledgments
a graph of connected facts by linking up docu-
The authors acknowledge the support by Airbus and the Royal
ments to pieces of information. The actualisation
Academy of Engineering under the Research Chairs and Senior
of uniform data interfaces and out-of-step data Research Fellowships scheme (RCSRF1718\5\41).
computing technologies is also a potential solu-
tion to these challenges.
Disclosure statement
7. Conclusion No potential conflict of interest was reported by the author(s).

This paper presents a literature analysis on the digital

twin concept to address the question: ‘How does the
digital twin concept support the realisation of an inte-
Funding
grated, flexible and collaborative manufacturing environ- Funding This work was supported by the Engineering and
ment as one of the goals projected by the fourth industrial Physical Sciences Research Council of the UK through the
revolution?’. A review of the literature was conducted to Future Electrical Machines Manufacturing Hub (EP/S018034/1)
investigate the development of the digital twin concept, and the Made Smarter Innovation -Research Centre for
Connected Factories (EP/V062123/1). The work was also sup-
maturity and its vital role within the manufacturing
ported by the Petroleum Technology Development Fund
industry. Six key functionalities inherent in the digital (PTDF) of the Federal Republic of Nigeria (PTDF/ED/PHD/OIU/
twin applications were identified: prognostic and diag- 1292/17).T
nostic analyses, simulation (online and offline), control,
monitoring/supervision and optimisation. The review
findings also highlighted that there was no common References
framework for a digital twin model creation. This study ACSE. 2018. “Festo Cyber-physical Smart Factory
proposed a conceptual framework that enables the inte- Announcement.” Accessed 20 March 2019. https://www.
gration of both product and process digital twin and the sheffield.ac.uk/acse/news/festophysicallab-1.794362
26 I. ONAJI ET AL.

Aivaliotis, P., K. Georgoulias, and K. Alexopoulos. 2019. Eisentrager, M., S. Adler, M. Kennel, and S. Moser. 2018.
“Using Digital Twin for Maintenance Applications in “Changeability in Engineering.” In 2018 IEEE International
Manufacturing: State of the Art and Gap Analysis.” In Conference on Engineering, Technology and Innovation
2019 IEEE International Conference on Engineering, (ICE/ITMC), 1–8. doi:10.1109/ICE.2018.8436295.
Technology and Innovation (ICE/ITMC), 1–5. doi:10.1109/ Ellgass, W., N. Holt, H. Saldana-Lemus, J. Richmond,
ICE.2019.8792613. A. Vatankhah Barenji, and G. Gonzalez-Badillo. 2018.
Bernard, A., and Z. Sandra. 2018. “Simulation and Virtual “A Digital Twin Concept for Manufacturing Systems.” In
Reality: Virtual Sensor Opens a World of Efficiency for Volume 2: Advanced Manufacturing. doi:10.1115/
Large Motors | Digital Twin | Siemens Global.” Accessed 27 IMECE2018-87737.
April 2021. https://new.siemens.com/global/en/company/ Enders, M. R., and N. Hoßbach. 2019. “Dimensions of Digital
stories/research-technologies/digitaltwin/virtual-sensor- Twin applications-A Literature Review.” In AMCIS 2019
opens-a-world-of-efficiency-for-large-motors.html Proceedings: Organizational Transformation & Information
Biesinger, F., D. Meike, B. Kraß, and M. Weyrich. 2019. “A Digital Systems (SIGORSA) 15-17, August, 2019 Cancun, Mexico.
Twin for Production Planning Based on Cyber-physical FEMM Hub Annual Report. 2021. “AnnuAl report 2021.” https://
Systems: A Case Study for A Cyber-Physical System-Based electricalmachineshub.ac.uk/wp-content/uploads/2021/04/
Creation of A Digital Twin.” Procedia CIRP 79: 355–360. FEMM-Hub-Annual-Report-2021.pdf
doi:10.1016/j.procir.2019.02.087. Grieves, M., and J. Vickers. 2016. “Digital Twin: Mitigating
Blum, M., and G. Schuh. 2017. “Towards a Data-oriented Unpredictable, Undesirable Emergent Behavior in Complex
Optimization of Manufacturing Processes.” Proceedings of the Systems.” In Transdisciplinary Perspectives on Complex
19th International Conference on Enterprise Information Systems Systems: New Findings and Approaches, 85–113. doi:10.
(ICEIS) 8 (1): 257–264. doi:10.5220/0006326002570264. 1007/978-3-319-38756-7_4.
Bohlin, R., J. Hagmar, K. Bengtsson, L. Lindkvist, J. S. Carlson, Haag, S., and R. Anderl. 2018. “Digital Twin – Proof of Concept.”
and R. Söderberg. 2017. “Data Flow and Communication Manufacturing Letters 15: 64–66. doi:10.1016/J.MFGLET.
Framework Supporting Digital Twin for Geometry 2018.02.006.
Assurance.” ASME 2017 International Mechanical Jones, D., C. Snider, A. Nassehi, J. Yon, and B. Hicks. 2020.
Engineering Congress and Exposition 2: V002T02A110– “Characterising the Digital Twin: A Systematic Literature
V002T02A110. doi:10.1115/IMECE2017-71405. Review.” CIRP Journal of Manufacturing Science and
Brenner, B., and V. Hummel. 2017. “Digital Twin as Enabler for an Technology 29: 36–52. doi:10.1016/j.cirpj.2020.02.002.
Innovative Digital Shopfloor Management System in the Esb Kabaldin, Y. G., D. A. Shatagin, M. S. Anosov, P. V. Kolchin, and
Logistics Learning Factory at Reutlingen - University.” Procedia A. M. Kuz’mishina. 2019. “CNC Machine Tools and Digital
Manufacturing 9: 198–205. doi:10.1016/J.PROMFG.2017.04.039. Twins.” Russian Engineering Research 39 (8): 637–644.
Cai, Y., B. Starly, P. Cohen, and Y.-S. Lee. 2017. “Sensor Data and doi:10.3103/S1068798X19080070.
Information Fusion to Construct Digital-twins Virtual Kritzinger, W., M. Karner, G. Traar, J. Henjes, and W. Sihn. 2018.
Machine Tools for Cyber-physical Manufacturing.” Procedia “Digital Twin in Manufacturing: A Categorical Literature
Manufacturing 10: 1031–1042. doi:10.1016/J. Review and Classification.” IFAC-PapersOnLine 51 (11):
PROMFG.2017.07.094. 1016–1022. doi:10.1016/j.ifacol.2018.08.474.
Catapult, H. V. M. 2018. “Feasibility of an Immersive Digital Kunath, M., and H. Winkler. 2018. “Integrating the Digital Twin
Twin.” https://www.amrc.co.uk/files/document/219/ of the Manufacturing System into a Decision Support
1536919984_HVM_CATAPULT_DIGITAL_TWIN_DL.pdf System for Improving the Order Management Process.”
Cheng, Y., Y. Zhang, P. Ji, W. Xu, Z. Zhou, and F. Tao. 2018. “Cyber- Procedia CIRP 72: 225–231. doi:10.1016/j.procir.2018.03.192.
physical Integration for Moving Digital Factories Forward Leng, J., H. Zhang, D. Yan, Q. Liu, X. Chen, and D. Zhang. 2019.
Towards Smart Manufacturing: A Survey.” The International “Digital Twin-driven Manufacturing Cyber-physical System
Journal of Advanced Manufacturing Technology 97 (1–4): for Parallel Controlling of Smart Workshop.” Journal of
1209–1221. doi:10.1007/s00170-018-2001-2. Ambient Intelligence and Humanized Computing 10 (3):
Cruz Salazar, L. A., D. Ryashentseva, A. Lüder, and B. Vogel-Heuser. 1155–1166. doi:10.1007/s12652-018-0881-5.
2019. “Cyber-physical Production Systems Architecture Based Liu, Z., N. Meyendorf, and N. Mrad. 2018. “The Role of Data
on Multi-agent’s Design Pattern—comparison of Selected Fusion in Predictive Maintenance Using Digital Twin.” AIP
Approaches Mapping Four Agent Patterns.” International Conference Proceedings 1949 (1): 20022–20023. doi:10.1063/
Journal of Advanced Manufacturing Technology 105 (9): 1.5034337.
4005–4034. doi:10.1007/s00170-019-03800-4. Lu, Y., C. Liu, -K. I.-K. Wang, H. Huang, and X. Xu. 2020. “Digital
DIN and DKE. 2018. “German Standardization Roadmap on Twin-driven Smart Manufacturing: Connotation, Reference
Industry 4.0.” Accessed 7 November 2019. https://www.din. Model, Applications and Research Issues.” Robotics and
de/en/innovation-and-research/industry-4-0/german- Computer-Integrated Manufacturing 61: 101837.
standardization-roadmap-on-industry-4-0-77392 doi:10.1016/j.rcim.2019.101837.
 INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING 27

Lu, Y., and X. Xu. 2018. “Resource Virtualization: A Core Towards Smart Manufacturing.” In Volume 1: Additive
Technology for Developing Cyber-physical Production Manufacturing; Bio and Sustainable Manufacturing.
Systems.” Journal of Manufacturing Systems 47: 128–140. doi:10.1115/MSEC2018-6435.
doi:10.1016/J.JMSY.2018.05.003. Qi, Q., F. Tao, Y. Zuo, and D. Zhao. 2018a. “Digital Twin Service
Lu, Y., X. Xu, and L. Wang. 2020. “Smart Manufacturing Process Towards Smart Manufacturing.” Procedia CIRP 72: 237–242.
and System Automation – A Critical Review of the Standards doi:10.1016/j.procir.2018.03.103.
and Envisioned Scenarios.” Journal of Manufacturing Systems Ríos, J., J. C. Hernandez-Matias, M. Oliva, and F. Mas. 2015.
56 (July 1): 312–325. doi:10.1016/j.jmsy.2020.06.010. “Product Avatar as Digital Counterpart of a Physical
Lyly-Yrjänäinen, J., J. Holmström, M. I. Johansson, and Individual Product: Literature Review and Implications in
P. Suomala. 2016. “Effects of Combining Product-centric an Aircraft.” Advances in Transdisciplinary Engineering
Control and Direct Digital Manufacturing: The Case of 657–666. doi:10.3233/978-1-61499-544-9-657.
Preparing Customized Hose Assembly Kits.” Computers in Rosen, R., G. Von Wichert, G. Lo, and K. D. Bettenhausen. 2015.
Industry 82: 82–94. doi:10.1016/j.compind.2016.05.009. “About the Importance of Autonomy and Digital Twins for
Macchi, M., I. Roda, E. Negri, and L. Fumagalli. 2018. “Exploring the Future of Manufacturing.” IFAC-PapersOnLine 48 (3):
the Role of Digital Twin for Asset Lifecycle Management.” 567–572. doi:10.1016/J.IFACOL.2015.06.141.
IFAC-PapersOnLine 51 (11): 790–795. doi:10.1016/J. Scaglioni, B., and G. Ferretti. 2018. “Towards Digital Twins
IFACOL.2018.08.415. through Object-oriented Modelling: A Machine Tool Case
Martinez, G. S., S. Sierla, T. Karhela, and V. Vyatkin. 2018. Study.” IFAC-PapersOnLine 51 (2): 613–618. doi:10.1016/J.
“Automatic Generation of a Simulation-based Digital Twin IFACOL.2018.03.104.
of an Industrial Process Plant.” In IECON 2018 - 44th Annual Schleich, B., N. Anwer, L. Mathieu, and S. Wartzack. 2017.
Conference of the IEEE Industrial Electronics Society, “Shaping the Digital Twin for Design and Production
3084–3089. doi:10.1109/IECON.2018.8591464. Engineering.” CIRP Annals 66 (1): 141–144. doi:10.1016/J.
Microsoft, C. 2017. “The Promise of a Digital Twin CIRP.2017.04.040.
Strategy.” https://info.microsoft.com/rs/157-GQE-382/ Schroeder G. N., Steinmetz C, Pereira C. E. and Espindola D.
images/Microsoft%27sDigitalTwin%27How-To% B. 2016. ”Digital Twin Data Modeling with AutomationML
27Whitepaper.pdf and a Communication Methodology for Data Exchange.”
Mitchell B and Yilmaz L. 2008. ”Symbiotic adaptive multisimu- IFAC-PapersOnLine, 49(30), 12–17. 10.1016/j.
lation.” ACM Trans. Model. Comput. Simul., 19(1), 1–31. ifacol.2016.11.115
doi:10.1145/1456645.1456647 Schuh, G., and M. Blum. 2016. “Design of a Data Structure for
Modoni, G. E., M. Sacco, and W. Terkaj. 2016. “A the Order Processing as a Basis for Data Analytics Methods.”
Telemetry-driven Approach to Simulate Data-intensive In 2016 Portland International Conference on Management
Manufacturing Processes.” Procedia CIRP 57: 281–285. of Engineering and Technology (PICMET), 2164–2169.
doi:10.1016/j.procir.2016.11.049. doi:10.1109/PICMET.2016.7806715.
Moher, D., A. Liberati, J. Tetzlaff, and D. G. Altman. 2009. “Preferred Shafto, M., M. C. Rich, D. E. Glaessgen, C. Kemp, J. Lemoigne,
Reporting Items for Systematic Reviews and Meta-analyses: and L. Wang. 2012. Modeling, Simulation, Information
The PRISMA Statement.” Journal of Clinical Epidemiology Technology & Processing Roadmap Technology Area 11.
62 (10): 1006–1012. doi:10.1016/j.jclinepi.2009.06.005. America: National Aeronautics and Space Administration
Nagabushanam, P., S. T. George, and S. Radha. 2020. “EEG (NASA). https://www.nasa.gov
Signal Classification Using LSTM and Improved Neural Shubenkova, K., A. Valiev, E. Mukhametdinov, V. Shepelev,
Network Algorithms.” Soft Computing 24: 1–23. S. Tsiulin, and K. H. Reinau. 2018. “Possibility of Digital
doi:10.1007/s00500-019-04515-0. Twins Technology for Improving Efficiency of the Branded
Negri, E., L. Fumagalli, C. Cimino, and M. Macchi. 2019. “FMU- Service System.” In Proceedings - 2018 Global Smart
supported Simulation for CPS Digital Twin.” Procedia Industry Conference, GloSIC 2018, 1–7. doi:10.1109/
Manufacturing 28: 201–206. doi:10.1016/j.promfg. GloSIC.2018.8570075.
2018.12.033. Sierla, S., V. Kyrki, P. Aarnio, and V. Vyatkin. 2018. “Automatic
Negri, E., L. Fumagalli, and M. Macchi. 2017. “A Review of the Assembly Planning Based on Digital Product Descriptions.”
Roles of Digital Twin in Cps-based Production Systems.” Computers in Industry 97: 34–46. doi:10.1016/J.COMPIND.
Procedia Manufacturing 11: 939–948. doi:10.1016/J. 2018.01.013.
PROMFG.2017.07.198. Stapic, Z., E. G. López, A. G. Cabot, L. de Marcos Ortega, and
Onaji, I., P. Soulatiantork, B. Song, D. Tiwari, and A. Tiwari. 2019. V. Strahonja. 2012. “Performing Systematic Literature
“Discrete Event Simulation for Cyber-physical System.” In Review in Software Engineering.” In Central European
Advances in Manufacturing Technology XXXIII: Proceedings of Conference on Information and Intelligent Systems
the 17th International Conference on Manufacturing Research, Varazdin, Croatia, 441.
edited by M. P. Y. Jin, 213–218. doi:10.3233/ATDE190037. Stark R, Fresemann C and Lindow K. 2019. ”Development and
Qi, Q., D. Zhao, T. W. Liao, and F. Tao. 2018b. “Modeling of operation of Digital Twins for technical systems and ser-
Cyber-physical Systems and Digital Twin Based on Edge vices.” CIRP Annals, 68 (1): 129–132. doi: 10.1016/j.
Computing, Fog Computing and Cloud Computing cirp.2019.04.024.
28 I. ONAJI ET AL.

Stark, R., S. Kind, and S. Neumeyer. 2017. “Innovations in Digital 2019 9th International Electric Drives Production
Modelling for Next Generation Manufacturing System Conference, EDPC 2019 - Proceedings. doi:10.1109/
Design.” CIRP Annals 66 (1): 169–172. doi:10.1016/J. EDPC48408.2019.9011857.
CIRP.2017.04.045. Weyer, S., T. Meyer, M. Ohmer, D. Gorecky, and D. Zühlke. 2016.
Talkhestani, B. A., N. Jazdi, W. Schlögl, and M. Weyrich. 2018. “Future Modeling and Simulation of CPS-based Factories: An
“A Concept in Synchronization of Virtual Production System Example from the Automotive Industry.” IFAC-PapersOnLine
with Real Factory Based on Anchor-point Method.” Procedia 49 (31): 97–102. doi:10.1016/J.IFACOL.2016.12.168.
CIRP 67: 13–17. doi:10.1016/J.PROCIR.2017.12.168. Yun, S., J.-H. Park, and W.-T. Kim. 2017. “Data-centric
Tao, F., F. Sui, A. Liu, Q. Qi, M. Zhang, B. Song, . . . A. Y. C. Nee. Middleware Based Digital Twin Platform for Dependable
2018. “Digital Twin-driven Product Design Framework.” Cyber-physical Systems.” In 2017 Ninth International
International Journal of Production Research 1–19. Conference on Ubiquitous and Future Networks (ICUFN),
doi:10.1080/00207543.2018.1443229. 922–926. doi:10.1109/ICUFN.2017.7993933.
Tao, F., H. Zhang, A. Liu, and A. Y. C. Nee. 2019. “Digital Zhang, C., G. Zhou, J. He, Z. Li, and W. Cheng. 2019a. “A Data-
Twin in Industry: State-of-the-Art.” IEEE Transactions on and Knowledge-driven Framework for Digital Twin
Industrial Informatics 15 (4): 2405–2415. doi:10.1109/ Manufacturing Cell.” Procedia CIRP 83: 345–350.
TII.2018.2873186. doi:10.1016/J.PROCIR.2019.04.084.
Tao, F., and M. Zhang. 2017. “Digital Twin Shop-Floor: Zhang, C., W. Xu, J. Liu, Z. Liu, Z. Zhou, and D. T. Pham. 2019b.
A New Shop-floor Paradigm Towards Smart “A Reconfigurable Modeling Approach for Digital
Manufacturing.” IEEE Access 5: 20418–20427. Twin-based Manufacturing System.” Procedia CIRP 83:
doi:10.1109/ACCESS.2017.2756069. 118–125. doi:10.1016/j.procir.2019.03.141.
Tao, F., M. Zhang, and A. Y. C. Nee. 2019. “Background and Zhang, D., S. Xu, S. Du, J. Wang, and J. Gong. 2017. “Progress of
Concept of Digital Twin.” In Digital Twin Driven Smart Pharmaceutical Continuous Crystallization.” Engineering
Manufacturing, 3–28. doi:10.1016/b978-0-12-817630- 3 (3): 354–364. doi:10.1016/J.ENG.2017.03.023.
6.00001-1. Zhang, H. Y., and J. Li. 2013. “Modeling Method and Application
Tharma, R., R. Winter, and M. Eigner. 2018. An Approach for in Digital Mockup System Towards Mechanical Product
the Implementation of the Digital Twin in the Automotive Pengcheng Wang, Xiangdong Liu and Yongquan Han.” In
Wiring Harness Field, 3023–3032. doi:10.21278/idc. Advanced Materials Research, Vol. 605, 604–608. Switzerland:
2018.0188. Trans Tech Publications.
Ward, R., P. Soulatiantork, S. Finneran, R. Hughes, and A. Tiwari. Zhang, H., L. Ma, J. Sun, H. Lin, and M. Thürer. 2019c. “Digital
2021. “Real-time Vision-based Multiple Object Tracking of Twin in Services and Industrial Product Service systems:
a Production Process: Industrial Digital Twin Case Study.” In Review and Analysis.” Procedia CIRP 83: 57–60. doi:10.1016/
Proceedings of the Institution of Mechanical Engineers, Part j.procir.2019.02.131.
B: Journal of Engineering Manufacture, 095440542110024. Zhou, G., C. Zhang, Z. Li, K. Ding, and C. Wang. 2019.
doi:10.1177/09544054211002464. “Knowledge-driven Digital Twin Manufacturing Cell Towards
Weber C, Königsberger J, Kassner L and Mitschang B. Intelligent Manufacturing.” International Journal of Production
(2017). M2DDM – A Maturity Model for Data-Driven Research 1–18. doi:10.1080/00207543.2019.1607978.
Manufacturing. Procedia CIRP, 63 173–178. doi: 10.1016/ Zhuang, C., J. Liu, and H. Xiong. 2018. “Digital Twin-based Smart
j.procir.2017.03.309 Production Management and Control Framework for the
Weigelt, M., J. Kink, A. Mayr, J. V. Lindenfels, A. Kuhl, and Complex Product Assembly Shop-floor.” The International
J. Franke. 2019. “Digital Twin of the Linear Winding Journal of Advanced Manufacturing Technology 96 (1–4):
Process Based on Explicit Finite Element Method.” In 1149–1163. doi:10.1007/s00170-018-1617-6.