Content uploaded by Keno Moenck
Author content
All content in this area was uploaded by Keno Moenck on Oct 07, 2022
Content may be subject to copyright.
33. DfX-Symposium 2022
© 2022 die Autoren | DOI: https://doi.org/10.35199/dfx2022.20
Digital Twins of existing long-living assets: reverse
instantiation of the mid-life twin
Keno Moenck1,*, Fabian Laukotka2, Dieter Krause2, Thorsten Schüppstuhl1
1 Institute of Aircraft Production Technology, Hamburg University of Technology
2 Institute of Product Development and Mechanical Engineering Design, Hamburg University of Technology
* Corresponding Author:
Keno Moenck
Hamburg University of Technology
Institute of Aircraft Production Technology
Denickestraße 17
21073 Hamburg
+49 (0)40 42878 3341
keno.moenck@tuhh.de
Abstract
The added value of long-living assets declines during their
lifespan, especially if they do not undergo regular planning-
intensive maintenance and retrofits. Here, the Digital Twin (DT)
concept can support by representing the physical assets most
recent state, typically based on data and information from
product creation. However, in the depicted domain, the
stakeholders of the products Mid-of-Life often do not have
access to the early phases. Therefore, as often presented in
current concepts, creating a holistic Digital Twin is not feasible.
Instead, in the Mid-Life phase of long-living assets, only a use-
case-specific and demand-actuated Digital Twin is attainable.
This instantiation requires a solid procedure, which will be
elaborated on in this work.
Keywords
digital twin, long-living assets, retrofit
2
1. Motivation
Long-living assets are often characterized by high value and complex structures on the
technological and system level, as well as multiple and changing involved stakeholders in roles
like the designer, manufacturer, owner, maintainer, or retrofitter. Therefore, continuous
Product Lifecycle Management (PLM) and its subordinate topic, Product Data
Management (PDM), face numerous data-intrinsic challenges such as the level of detail and
model fidelity but also stakeholder-interrelated ones like accessibility or confidentiality [1].
Maintaining highly sophisticated and ever-changing systems or integrating new
components requires solid planning and engineering efforts in the Product Usage phase.
These tasks rely on a comprehensive digital as-is data basis, also known as the Digital
Twin (DT), contrary to the available data and information in practice. Especially for long-living
assets, the Mid-Life stakeholders usually do not have sufficient access to information
from the early phases caused, e.g., by market-based segregation of the OEM and subsequent
service provider or by data that originated before the current rise of the importance of digital
data and the technological ability to handle and also exchange it [2].
Typically, DTs are based on knowledge and information derived from the early
lifecycle phase that must be constantly updated with data, often sensor-based acquired, to
form the as-is or current state [3]. Therefore, non-continuities in the data flow, thus
missing necessary information in the Mid-Life, require a retrospective acquisition and
creation. In this work, the result of this procedure is called the Mid-Life-Twin (MLT), as depicted
in Figure 1. This work will elaborate on an approach, a process model, to instantiate this Digital
Twin. The main objective is not to establish the MLT as a new term but to indicate the deviation
between the Digital Twin concepts commonly defined in the literature (s. Section 2.2) and the
practice of instantiating DTs for long-living assets in different domains, independent from the
product creation.
Besides the usage, objectives in a Mid-of-Life (MoL) are usually the
life extension and preservation. Some strategies include replacing, refurbishing,
maintaining, or de-rating [4,5]. In contrast, the activity retrofit updates the asset with new
technology and innovations, so essential to keep a long-living asset profitable and
sustainable benefiting not only the owner but also the resource-limited environment. Out of
the MoL activities, retrofit usually requires the most comprehensive digital replica of the asset;
therefore, this scope is constrained to this activity.
Figure 1: The use case and challenges of the Mid-Life-Twin (MLT) (derived from [6]).
3
This work is structured as follows: First, in Section 2, related work is presented, covering
retrofit examples in different domains, referencing the standard Digital Twin definition and a
common process model in Data Science used to frame the subsequently presented
approach (s. Section 4). In Section 5, the derived process model is applied to an example use
case in order to demonstrate its capabilities.
2. Related Work
2.1. Retrofit of Long-Living Assets
Retrofit as an assets life extension driver, increasing its added value or decreasing its
environmental emissions, is applied in different domains, but primarily for long-living assets
where the sub-systems technology evolves faster than the asset itself and modifications are
more economical than the disposition or rebuilding from scratch.
An example of the last driver is building retrofit with objectives like introducing energy-saving
technical measures, e.g., increasing heat insulation performance [7]. Digitizing, managing, and
using a buildings data in its complete lifecycle is subject to Building Information Modeling
(BIM), similarly to the concept of PLM. Since, especially for long-existing buildings, not all
necessary data is digitally available, data acquisition by, e.g., laser scanning, is a core
challenge in as-built BIM [8] that also includes generating model-based and linked information
based on the 3D scans [9]. The data and information gap in BIM for new construction projects
is far less than for existing buildings due to the recent awareness of continuous and
collaborative data management. For existing buildings, technical, cost, organizational, and
legal challenges or missing or obsolete data inhibit a consistent BIM [10].
The need to acquire the as-built environment in production plants or factories is closely
related to the previous examples. Similar to buildings, all these assets have their unique history
and lifecycle, where the creation of an up-to-date digital production system twin enables
various planning processes digitally, like retrofit and refurbishment [11]. as-isor as-build
(terminology depends on the domain and the point in time of expected changes) data collection
is used in different domains in the context of retrofitting. Examples are, e.g., the naval [12],
offshore [13,14], and aircraft [15,16] industry, using 3D scanning technology and subsequent
processing strategies. Besides the sensory-based data acquisition, in th
proposition (Sec. 4), an overview of data collection methods to bridge the outlined gap of
missing data between an assets initial and later in-life phase will be presented.
2.2. Digital Product Twin
The CIRP Encyclopedia of Production Engineering defines the Digital Twin (DT) as a digital
phases, where the product is either a tangible/intangible asset or a service [17]. In practice,
the holistic, interconnected representation of the complete product or asset is not necessary
or feasible. Digital Twins serve purposes and thus must only represent the needed component
and (sub-)system [18]. According to the previously cited references and common
understanding found in literature, the DT is based on models (Digital Master) from the products
development (BOL) [3], as depicted in Figure 2. Thus, there is an intrinsic need for a Digital
Master to instantiate a Digital Twin for the respective asset.
4
Figure 2: The Digital Twin Concept (based on the illustration by [3]).
Contrary to this continuous concept in terms of the life phases, within the presented scenario
of this paper, the information from the Engineering phase (red arrows) at the Begin-of-
Life (BoL) does not exist and, thus, the Digital Master needs to be created differently. Previous
Section 2.1 outlines examples from different domains, where data collection and subsequent
data modeling are one of the main tasks in deriving and DT, also
termed Twinning. For this activity, in Section 4, a process model from analysis to collection
and data preparation is presented.
2.3. Learning from Data Science
Due to the challenge regarding the data collection and processing the scenario in this work
contains, a view on domains frequently facing similar challenges stands to reason. Subject to
Data Science is handling data in high volume, e.g., datasets containing millions of entries,
using defined concepts and procedures. Usually, data is categorized as structured, semi-
structured, or unstructured data [19] and characterized by the 5 Vs of Big Data (volume, variety,
velocity, validity, veracity, and value) [20].
One of the most applied process models in Data Science is the CRoss-Industry Standard
Process for Data Mining (CRISP-DM), that encourages best practices application- and
industry-neutral, dividing the data mining process into six phases: business understanding,
data understanding, data preparation, modeling, evaluation, and deployment [21]. Within these
six phases (s. Figure 3), the model proposes different subordinate steps, e.g., Determining
Business Objectives or Collect Initial Data, that the operator performs data- and demand-
specific. The model defines a best practice on a structural level; the specific formulation of
actions executed in a single work step and task is up to the user. The model advises what
needs to be incorporated in general. That also allows for the addition of custom steps and
extension of given tasks and procedures. Therefore, CRISP-DM is broadly accepted and
implemented worldwide [22]. Intentionally, the process model refers to the steps of data mining,
extracting information, and discovering patterns in large datasets. Section 4 will show that the
model also suits our proposed approach due to its flexibility to fit various analytic tasks.
Figure 3: The CRISP-DM process model [21].
5
3. Methodical Approach
A representative and field of application for the depicted scenario were identified within
aviation. The basic situation was analyzed more closely in cooperation with a leading
maintenance and retrofit company. Especially the availability and necessary data and
information to plan and perform aircraft-cabin conversions were considered. In Section 5, this
use case is presented. An elementary part of the presented work was to frame the process
with CRISP-DM into an overarching process model feasible to be established and form a
baseline adaptable to other Digital Twin use cases.
4. Process Model
This
MoL (mostly) independently from its BoL phase. There are primarily two DT design concepts:
data-based and system-based [23]. The first focuses on data and its modeling, e.g., in IoT
platforms like PTC ThingWorx [24], with the primary objective to structure and link sensory-
based gathered information in data models without extensively modeling the system
characteristics. In contrast, the system-based design process starts with modeling the system
of interest at, e.g., its logical, physical, or functional level. Herefore, an engineer needs a much
more in-depth technical insight and understanding [23]. Besides isolated system- or data-
driven modeling, combinations and linkages are required depending on the
extent of the purposes [16].
Since the data-driven approach is usually top-down (from data to the insight),
compared to the system-driven design process (bottom-up), and the pre-requisite of missing
knowledge from the BoL, the process model follows the data-driven design approach.
However, it indeed allows for later system or discipline-specific model integration.
In order to follow the data-driven approach, the procedure is framed by utilizing
CRISP-DM (s. Section 2.3) and its six phases as a superordinate procedure to derive the
Digital Master and Digital Shadow (s. Figure 4). The individual tasks will be formulated and
extended in the following sections while focusing on the specific challenges of the Digital Twins
reverse-instantiation, identifying required but missing information, and acquiring additional
data using, e.g., re-engineering processes.
Figure 4: Process model to instantiate the MLT (highlighted in green: changes to the CRISP-DM model [21]).
4.1. Business Understanding
The Business Understanding phase is the upstream point aiming to define the overall project
from a business perspective up to Producing a Project Plan. In the steps, Determine
Business Objectives and Goals, the value-adding application(s) must be defined while
constraining the scope to specific achievable goals based on the application(s). An application
may aim at monitoring, simulation, prediction, or verification [25]. For the chosen application,
a discipline expert must then define the necessary models and needed data while also
6
Assessing the Situation, which means gathering everything by inventorying available data
resources to software. Without looking into the data, the goal is to determine everything already
at hand.
4.2. Data Understanding
With the start of the second phase, Data Understanding, data will be accessed for the first
time. Closing this phase, everything needed is available, although the data might still be
unwrought. CRISP-DM includes four steps starting with the Collection of the Initial Data and
ending with the Verification of the Data Quality. In between, the collected and loaded data
will be analyzed during the steps Describing and Exploring. At this point, all easily accessible
data sources have been collected, analyzed, and verified. In the context of the MLT, the
presented procedure extends CRISP-DM by adding two subsequent steps: Analyze
Completeness and consequently Acquisition of Additional Data. Instead of integrating
these in the other Data Understanding steps, they are classified as novel to emphasize the
subsequent execution to prevent redundant data acquisition. Since the original Data
Understanding steps are straightforward to adapt, the following scope is constrained to the two
newly added ones.
4.3. Analyze Completeness
The user defines in the Determine Goal step all required resources for the necessary
discipline-specific models, e.g., a geometric Boundary Representation (B-rep) or a logical
control model (s. [26] for further examples), b n. All
accessible data were collected and analyzed within the previous steps after identifying the
easily available data silos. In this phase, the user must dive into the data itself and assess
whether all needed information can actually be drawn from the given resources. The user can
continue with phase 3 of the process model if all necessary information is already available.
Otherwise, the identified gap must be closed by proceeding to step (2.d.) the acquisition of
the missing data and, thus, the main task required for the reverse instantiation.
4.4. Acquire Additional Data
While some information may already be available, e.g., from previous projects or
applications, there is a need to acquire additional data. This phase can be categorized into
three main groups: external collection, manual and sensory-based acquisition. However, how
the additional information is acquired depends on the data and the new resources that can be
identified and accessed.
Some information may be easily acquired in terms of cost and effort by requesting them
from external stakeholders, like manufacturers or suppliers, e.g., delivered in the form of
unstructured textual data or even modeled ones like a CAD file. However, in the case of long-
living assets, data and information often may not be accessible, either because of intellectual
property regulations or more apparent ones, e.g., the original suppliers are not enterprising
anymore.
If the asset is accessible, some information may be acquired manually, e.g., visually or
using simple measuring devices. Besides much other information, simple geometric
information or a Bill of Materials (BOM), ideally with serial numbers, can be acquired this way.
Any operator with physical access to the relevant area of the asset can generally perform these
manual activities with or without the need to take it out of operation.
7
If data of more volume, velocity, or variety is necessary, sensory-based acquisition, e.g.,
using 3D-scanning technologies, can be a data provider. However, access to the asset for a
longer timespan and probable taking it out of operation is required, as the acquisition or sensor
installation processes take time. Additionally, this acquisition is not finished with the scan itself,
but the data needs to be processed further. The VDI 5620 Reverse Engineering of
Geometrical Data [27] is a guide for performing this task as it also considers factors like
accuracy, processes, and required hardware; recommended if the operator requires
deriving, e.g.,
While the acquisition strategy depends on the application, model to derive for the Digital
Master, type of data, and accessible sources, as introduced in Section 2.3, data science
distinguishes between structured-, semi-structured, and unstructured data. This taxonomy can
guide a discipline expert in choosing an acquisition strategy depending on the type of data. A
generic overview of acquisition strategies for different domains and their respective constraints
Instead, an example overview showing a selection of acquisition
strategies in the example use case is depicted in Figure 5. Eventually, with additional data
acquired, a loop back to (2.b.) in the process model is advised to ensure once more that
everything needed is available and sufficient to realize the application.
4.5. CRISP-DM Steps 3-6
With the beginning of phase 3, all needed data is available, maybe still unwrought, without
context, and not model-based. Phase 3 (Data Preparation) includes activities to select the final
set of data, filter unnecessary, create data context to gain information, and integrate as well as
format everything [21]. For example, the operator may extract the necessary assembly
modules from CAD files and export them into another file format. Also, phase 3 will, e.g.,
include steps of the previously mentioned VDI 5620: postprocessing, data fusion, and
registration. Unstructured or semi-structured data need context to derive information and
create the necessary data models. For example, textual information from ill-posed modeled
files, e.g., PDF files, must be extracted and stored in data models embedded in databases.
In phase 4 (Modeling), this now well-formatted and organized information can be used to
actually create the required cohesive model representing the digital asset with all its needed
facets the Digital Twin is being instantiated. The models form the Digital Master, while the
Digital Shadow consists of
digital instance. Overall, phase 4 includes the Selection of the Modeling Technique,
Building the Model, and Assessing the Model. As a guide in choosing a tool, Qi et al. give
a comprehensive overview of enabling tools for these activities, including various software
capable of geometric, physical, rule, or behavioral modeling [25].
Within the scope of the original CRISP-DM, the next phase is the Evaluation of the resulting
model facing the overall objectives. In this scenario, this is translated into evaluating whether
the resulting model includes all information defined
during the Business Understanding phase. In case of failures, loopbacks to the first
phase (Business Understanding) may result.
Formalized following CRISP-DM, the last phase describes the Deployment
applications and services. The tools will depend on the model types and connection capabilities
between the physical-digital or digital-digital spaces (s. [25] for an excerpt on market-available
software).
5. Example Use Case
An example use case for the presented Mid-Life-Twin is the retrofit of an aircraft, during
which a part of the cabin shall be updated with new equipment. In order to reliably plan the
new cabin layout errorless, minimizing aircraft ground times, an as-is digital representation of
8
the aircraft is required. In the scenario, a third-party company performs the retrofit, as usual in
this industry, and the last considerable modification are
more than a decade ago. Therefore, a comprehensive digital representation of the specific
aircraft is missing. The aircraft operator plans to perform all following modifications in
cooperation with one selected retrofit company. The instantiation of the MLT is intended to
allow for easier planning of the current and future modifications and maintenance works.
The use case and twin application are defined within phase (1) of the presented process
model. Based on prior modifications, the needed models and information are identified: mainly
geometric details and information about the relevant aircraft system interfaces of the specific
instance, especially in the area near the modification. Then, it is identified that the retrofit
organization already has some generic information about the type of aircraft available (1.b.-c.).
During the first steps of phase 2, the available information is analyzed, resulting in a general
overview of the type of an aircrafts critical geometries and dimensions. However, it is also
identified that the specific aircraft does not match the generic descriptions because of prior
modifications and maintenance (2.b.). Thus, more information about the assets actual state is
required before a representative geometrical and descriptive model can be created.
Within the step of Analyze Completeness (2.c.), a list of the needed information to plan
the new cabin and modification is derived and categorized by data type. Subsequently, in
step (2.d.), different strategies are used to acquire mentioned information. A representative
overview of acquisition strategies is depicted in Figure 5.
Figure 5: Overview of selected example acquisition strategies used in the described use case.
After the data and information are acquired using different strategies, everything is analyzed
to ensure sufficient quality and all required resources are available. At this point, the
application-specific reverse-instantiation is started by following the subsequent phases and
combining all available information into an aircraft-specific representation the Digital Twin.
Then, data and information are made available to engineers using state-of-the-art PLM
software. These can also update the model with new data based on the planned modification.
However, the documentation of interdependencies between data and the aircraft schematics
like known structure can be further improved, e.g., by using system models, as already
presented in past related work [16].
9
6. Conclusion
In contrast to common Digital Twin definitions in the literature, in the case of long-living
assets, often a third party needs to instantiate and then update the digital counterpart
without much prior data and information. Then, data handling, one of the most crucial aspects
of Digital Twins, becomes even more challenging since the presence of a single data and
information source is unlikely. This work outlines the data and information gap in this scope
and presents a procedure framed by the CRISP-DM process model originating from Data
Science. So, the reverse instantiation of a Digital Twin in the concept of the Mid-Life-Twin is
the retroactive reconstruction of the necessary information to create the digital representation
of the physical asset. In essence, this includes the analysis of the needed data for the retrofit,
analyzing and collecting information available from general knowledge or the product creation,
and identifying mismatches and differences between these two tasks to select appropriate
methods to fill this gap. With an example at hand, the two new steps in the Data Understanding
phase were discussed in more detail. Future work might focus more and elaborate on phases
3 to 6 and their individual steps. However, a more detailed distinction between domains and
their specific applications will be inevitable.
As in most publications about Digital Twin applications, generic or overarching process
models could be further developed and outlined to create, instantiate, and update a Digital
Twin methodically. As CRISP-DM is a guideline to encourage best practices to solve problems
in Data Science, a best practice in the form of a process model to assist the creation,
instantiation, constant update, and usage of Digital Twins would benefit the industry and
research community. This work presented an approach to a process model for creating and
instantiating the Mid-Life-Twin.
In addition, many software tools on the market enable various functional Digital Twin
aspects. However, the various applications make choosing or combining the right ones
challenging for any operator. Hence, considering software selection guidelines in these
process models is suggested. As the reverse instantiated Digital Twin is based on a variety of
single information, the documentation and usage of its metadata like interdependencies could
prospectively be improved using system modeling techniques.
Acknowledgment
This work is part of the research project Intelligent Digital Cabin Twin (InDiCaT), supported
by the Federal Ministry for Economic Affairs and Climate Action as part of the Federal
Aeronautical Research Programme LuFo VI-1.
Literature
[1] Stark, John: Product Lifecycle Management (Volume 2): The Devil is in the Details. 3rd ed. Cham: Springer
International Publishing, 2015 (Decision Engineering Ser). DOI: 10.1007/978-3-319-24436-5.
[2] Stark, Rainer; Grosser, Hendrik; Müller, Patrick: Product analysis automation for digital MRO based on
intelligent 3D data acquisition. In: CIRP Annals 62 (2013), Nr. 1, S. 123126.
DOI: 10.1016/j.cirp.2013.03.079.
10
[3] Stark, Rainer; Anderl, Reiner; Thoben, Klaus-Dieter; Wartzack, Sandro: WiGeP-
. 4750. DOI: 10.3139/104.112311.
[4] Morey, Stephen; Chattopadhyay, Gopinath; Larkins, Jo-ann: Decision-making in complex asset life
extension. In: 2021 International Conference on Maintenance and Intelligent Asset Management (ICMIAM):
IEEE, 2021 - 2021, S. 16. DOI: 10.1109/ICMIAM54662.2021.9715202.
[5] Roy, R. et al.: Continuous maintenance and the future Foundations and technological challenges. In: CIRP
Annals 65 (2016), Nr. 2, S. 667688. DOI: 10.1016/j.cirp.2016.06.006.
[6] Laukotka, Fabian; Hanna, Michael; Schwede, Lea-Nadine; Krause, Dieter: Lebensphasenübergreifende
Nutzung Digitaler Zwillinge. In: Zeitschrift für wirtschaftlichen Fabrikbetrieb 115 (2020), s1, S. 101104.
DOI: 10.3139/104.112332.
[7] Bu, Shanshan et al.: Literature review of green retrofit design for commercial buildings with BIM implication.
In: Smart and Sustainable Built Environment 4 (2015), Nr. 2, S. 188214.
DOI: 10.1108/SASBE-08-2014-0043.
[8] Sanhudo, Luís et al.: Building information modeling for energy retrofitting A review. In: Renewable and
Sustainable Energy Reviews 89 (2018), S. 249260. DOI: 10.1016/j.rser.2018.03.064.
[9] Sadeghineko, Farhad; Kumar, Bimal: Development of Semantically Rich 3D Retrofit Models. In: Journal of
Computing in Civil Engineering 34 (2020), Nr. 6. DOI: 10.1061/(ASCE)CP.1943-5487.0000919.
[10] DAngelo, Letizia; Hajdukiewicz, Magdalena; Seri, Federico; Keane, Marcus M.: A novel BIM-based process
workflow for building retrofit. In: Journal of Building Engineering 50 (2022), S. 104163.
DOI: 10.1016/j.jobe.2022.104163.
[11]
Optimization in a Built Environment. Cham: Springer International Publishing, 2022 (Springer Series in
Advanced Manufacturing). DOI: 10.1007/978-3-030-77539-1.
[12] Rata, Vasile; Secobeanu, Steluta: Aspects of using 3-D laser scanning technology in ship retrofit projects. In:
Annals of Dunarea de Jos University of Galati 42 (2019), S. 157162.
DOI: 10.35219/AnnUGalShipBuilding.2019.42.21.
[13] Anderson, Stephen; Barvik, Sigve: Advanced Offshore Digital Inspection Methods. In: International
Petroleum Technology Conference, 2020. DOI: 10.2523/IPTC-20087-Abstract
[14] Hokstad, Per; Håbrekke, Solfrid; Johnsen, Roy; Sangesland, Sigbjørn: Ageing and life extension for offshore
facilities in general and for specific systems : Report. March 2010.
[15] Deneke, Constantin; Oltmann, Jan; Schüppstuhl, Thorsten; Krause, Dieter: Technology Innovations for
Faster Aircraft Cabin Conversion. In: Estorf, Otto von; Thielecke, Frank (Hrsg.): 7th International Workshop
on Aircraft System Technologies, AST 2019, 2019.
[16] Moenck, Keno et al.: Towards an Intelligent Digital Cabin Twin to Support an Aircrafts Retrofit and Base
Maintenance. In: SAE Technical Paper Series. DOI: 10.4271/2022-01-0046.
[17] Stark, Rainer; Damerau, Thomas: Digital Twin. In: CIRP Encyclopedia of Production Engineering. Berlin,
Heidelberg: Springer Berlin Heidelberg, 2020, S. 18. DOI: 10.1007/978-3-642-35950-7_16870-1.
[18] Wilking, Fabian; Schleich, Benjamin; Wartzack, Sandro: Digital twins - Definitions, classes and business
scenarios for different industry sectors. In: Proceedings of the Design Society 1 (2021), S. 12931302.
DOI: 10.1017/pds.2021.129.
[19] Furht, Borko; Escalante, Armando: Handbook of Data Intensive Computing. Springer New York, 2011,
DOI: 10.1007/978-1-4614-1415-5.
[20] Matzka, Stephan: Künstliche Intelligenz in den Ingenieurwissenschaften. Springer Fachmedien Wiesbaden,
2021, DOI: 10.1007/978-3-658-34641-6.
[21] Shearer, C.: The CRISP-DM model: the new blueprint for data mining. In: Journal of Data Warehousing 5
(2000), Nr. 4, S. 1322.
[22] Huber, Steffen; Wiemer, Hajo; Schneider, Dorothea; Ihlenfeldt, Steffen: DMME: Data mining methodology for
engineering applications a holistic extension to the CRISP-DM model. In: Procedia CIRP 79 (2019),
S. 403408. DOI: 10.1016/j.procir.2019.02.106.
[23] Adamenko, Dmytro et al.: Review and comparison of the methods of designing the Digital Twin. In: Procedia
CIRP 91 (2020), S. 2732. DOI: 10.1016/j.procir.2020.02.146.
[24] PTC ThingWorx: Accelerate Success With ThingWorx IIoT Solutions Platform. Online:
https://www.ptc.com/en/products/thingworx. Accessed on: 02.08.2022.
[25] Qi, Qinglin et al.: Enabling technologies and tools for digital twin.
In: Journal of Manufacturing Systems 58 (2021), S. 321. DOI: 10.1016/j.jmsy.2019.10.001.
[26] Pasquinelli, Mauro; Molina-Tanco, Luis; Reyes-Lecuona, Arcadio; Cencetti, Michele: Extending the System
Model. In: Dynamics of long-life assets: From technology adaptation to upgrading the business model.
Springer Open, 2017, S. 169189. DOI: 10.1007/978-3-319-45438-2_10.
[27] VDI 5620. 2017-03. Reverse engineering of geometrical data.