The Digital Twin Theory - A New View on a Buzzword
München, 18. März 2019
Prof. Dr. rer. nat. Dipl.-Ing. Andreas Deuter has been a professor of computer science for engineering and production within the Department of Production and Economics at OWL University of Applied Sciences since 2015. His research work focuses on software development processes in the manufacturing industry, specializing in optimizing their integration in a higher-level product lifecycle.
Florian Pethig, M. Sc., is an IT engineer and has been working on solutions for data acquisition and management in automation at the Fraunhofer IOSB-INA institute since 2011. Since 2017, he has been leading a research group on big data platforms at the institute. His research work focuses on interoperable information models and communication for Industry 4.0. He is the main author of a VDMA guideline on Industry 4.0 communication and is the chairman of the I4AAS working group, a joint initiative between the OPC Foundation, ZVEI, and VDMA. He is also an active member of the Industry 4.0 platform (GMA 7.20).
The digital twin is seen as a major tool for increasing productivity in the age of industrial digitalization. A number of publications are therefore focusing on this concept. This article aims, first of all, to reveal the origins of the term and to discuss a selection of definitions. However, these are of little value in the practical implementation of digital twins, as the definitions vary greatly in some cases. Accordingly, a theoretical model which incorporates assumptions about the digital twin is proposed as an alternative to a classic definition. This novel approach aims to help improve the management of digital twins in practical application.
A digitalized industrial sector offers huge economic potential: In the mechanical and plant engineering sector alone, a cumulative increase in productivity of 30% is anticipated by 2025 as a result of Industry 4.0 . This increase is based in essence on the seamless networking of all stakeholders and systems both horizontally and vertically.
Various research projects are already dedicated to improving the vertical networking of the hierarchy levels in line with IEC 62264 as well as the procedures and services based on this. Examples include projects on predictive maintenance based on sensor data . In addition, work is ongoing to find solutions to the unfavorable situation regarding the vertical integration of information which is caused by the use of many different fieldbuses and IT protocols. For example, a comprehensive Industry 4.0 communication method on the basis of standardized information models is currently being developed .
The need for digital twins
In the future, machines and plants will be able to be incorporated into higher-level systems via “plug & monitor” without the high degree of integration work that is currently required. However, vertical networking alone is not enough to achieve the sought-after increases in productivity.
The next step that is required is to enhance the horizontal networking of the value chains. This horizontal networking is described in the Reference Architectural Model Industry 4.0 (RAMI 4.0) of IEC 62890 in the lifecycle & value stream axis . Along this axis, which represents aspects including development, production, and use in the product lifecycle, practitioners currently have to work with proprietary interfaces and information models . Structural information and models from design and engineering tools may make it easier to diagnose faults in production machines, for example, but such information and models are not currently compatible with the systems the machines use. As a result, the greatest increase in productivity in this sector is expected to be seen in cost-intensive engineering . The digital twin has an important role to play here. According to Gartner, it is a key element of product lifecycle management that has the potential to save several billion euros .
In addition to the lifecycle & value stream axis, the digital twin also plays a role in the layers axis and in the hierarchy levels axis of RAMI 4.0. Figure 1 presents a graph of the RAMI 4.0 space in which the digital twin “floats”. However, these types of images do not help practitioners clarify how the hoped-for increases in productivity are to be achieved. Therefore, we will first of all take a more in-depth look at the current definitions of the digital twin.
Figure 1: Digital twin in the context of RAMI 4.0.
In the technical domain, the concept of a twin was established by NASA in the late 1960s. This referred to the identical reproduction of a spacecraft that remained on earth to analyze the effects of control commands before sending them to the remote spacecraft. It was again NASA that added the attribute “digital” to a technical twin for the first time in 2010. In this case, they were referring to a simulation model that mapped the behavior of a physical spacecraft .
Around the same time, the term also appeared in the industrial domain. Here, it referred to a virtual copy of a physical product in product lifecycle management (PLM) systems . However, the term first became popular with the emergence of the idea of Industry 4.0 and when companies began to use the digital twin for their own marketing purposes, such as in . Since then, numerous definitions have arisen, as the following selection shows:
The digital twin is a digital representation of things from the real world ; a concept with which data and information of atoms is assigned to bits ; a computer-aided model of a tangible or intangible object ; a comprehensive physical and functional description of a product which includes all information to process it ; a digitalized (3D) reproduction of a product to be created ; a synonym for the Industry 4.0 asset administration shell .
As a systematic mapping study conducted as part of a project paper in accordance with the approach described in  reveals, the list of definitions could be extended further. The study, which so far only considers English-language articles in the ACM Digital Library, Science Direct, and IEEE Xplore databases, lists 51 relevant publications in which the digital twin is defined. What’s more, further similar terms such as digital shadow, digital master, digital type, and digital instance are also in circulation.
The following interim conclusion can therefore be drawn: There are a large number of definitions which vary in terms of their scope, degree of detail, and technical focus. In the main, the digital twin is understood to be a simulation model with a defined form; however, this is not universally valid or accepted. In principle, the existence of various definitions for a scientific subject do not prevent it from being implemented. In reality, however, there are numerous challenges when it comes to the management of digital twins, such as the identification and data management of the product along the product lifecycle, the creation of simulation models in different IT systems, and the management of huge data volumes . Formulating a precise definition of the term and working to get this generally accepted would be one way to attempt to overcome these challenges. However, we are proposing the discussion of an alternative approach, as we do not think it is realistic to expect the many stakeholders from science and industry to agree on one definition.
The digital twin theory
Figure 2: Model of information enrichment for digital twins (according to ).
This approach is a theoretical model based on hypotheses. The starting point for the hypotheses was firstly the work in  which states that the information that describes a digital twin is enriched in each phase of the product lifecycle (Figure 2). Secondly, chance contact with quantum physics and the topic of electrons led to the idea of the digital twin theory: From the perspective of quantum physics, electrons are located in several places simultaneously. Their state is unknown until they are moved to a monitoring state. It appeared interesting to investigate whether these characteristics could also be assumed for digital twins.
Following the initial formulation, the hypotheses were discussed with industry representatives including with a professional forum held by engineering association OWL Maschinenbau e.V. in June 2018  and at the PLM Europe conference in October 2018 . They were then revised and reformulated. The hypotheses of the digital twin theory are as follows:
- A digital twin is a digital representation of an asset.
- A digital twin is located in several places simultaneously.
- A digital twin has multiple states.
- The digital twin has a context-specific state in a specific interaction situation.
- The information model for digital twins is infinitely large; it is a real information model.
- The real information model can be finitely approximated for a specific application scenario, becoming a rational information model.
- The rational information model cannot be stored in a single place.
- The rational information model is never completely visible.
Figure 3 explains these hypotheses. An asset is an object of value. What an asset is precisely for a specific application scenario depends on the application scenario. It does not matter whether this object is tangible or intangible, a product or production system, a type or an instance. The digital twin is visible along the product lifecycle at several places and interacts at these places with an actor (person, machine, etc.). Consequently, the digital twin has multiple states. However, the digital twin is moved to a context-specific state in a specific interaction situation. One example of an interaction situation is the creation of the CAD model for a product type (context) by an engineer (actor). In this case, a CAD model has a state (in progress, etc.). Another example of a specific interaction with the digital representation of the same asset, in this case the product type, is the software design (context) by a software architect (actor).
The information describing a digital twin is therefore very different and depends on the asset. It is therefore not possible to define a complete information model for digital twins. The information model is infinitely large and is to be understood as a real information model. The attribute “real” comes from mathematics where the realm of real numbers comprises rational and irrational numbers. However, in order to be able to interact with a digital twin in a specific application scenario, an approximated information model must exist. We refer to this as the rational information model, another term which is derived from mathematics. As can be seen in Figure 3, the data for the rational information model is scattered along the product lifecycle. It is not saved in one place, for example, in a central database. In order to supply the data required for a specific interaction situation to a specific actor, this data must be transported via an appropriate interface infrastructure. This means that all the data from the rational information model is never completely visible.
What conclusions should be drawn
Figure 3: Possible infrastructure for digital twins.
Increases in productivity can be achieved through the digitalization of products and production. The vertical integration of factory and IT systems is making great strides in this regard. However, horizontal integration across the product lifecycle by means of digital twins offers at least as much potential for achieving increases in productivity, particularly in the field of the engineering. Despite this, digital twins are not yet clearly defined, making it difficult to manage them in practice. This article proposes a theoretical model to move away from attempts to establish a clear definition and instead to concentrate on specific mechanisms and added value of the abstract term.
As a scientific theory can only be disproved rather than proved, analysis of the hypotheses presented above is required. This has not yet been carried out, as the aim of this article is to introduce the idea of digital twin theory and to present it for discussion as an alternative to a classic definition. In order to define digital twin theory more precisely, active debate of the hypotheses it contains is required in further research work. Leading activities in this regard include the research project “Technical Infrastructure for Digital Twins” which has been initiated by the Fraunhofer IOSB-INA institute and OWL University of Applied Sciences as part of the “it’s OWL” leading-edge cluster .
 Bauer, W.; Ganschar, O.: Industrie 4.0 - Volkswirtschaftliches Potenzial für Deutschland, BITKOM-Studie, 2014
 Siemens provides online condition monitoring for predictive maintenance for NASA at the Armstrong Flight Center, https://news.usa.siemens.biz/press-release/condition-monitoring/siemens-provides-online-condition-monitoring-predictive-maintenan, Accessed 20.11.2018
 Jasperneite, J.; Niggemann, O.: Industrie 4.0-Kommunikation auf Basis von OPC UA – Leitfaden für die Einführung in den Mittelstand, VDMA Verlag, 2017.
 DIN SPEC 91345: Referenzarchitekturmodell Industrie 4.0 (RAMI4.0), Deutsches Institut für Normung (DIN) e.V., 2016
 Wagner, C.; Grothoff, J.; Epple, U.; Drath, R., Somayeh, M.; Grüner, S.; Hoffmeister, M.; Zimermann, P. : The role of the Industry 4.0 asset administration shell and the digital twin during the life cycle of a plant, IEEE International Conference on Emerging Technologies and Factory Automation (ETFA), 2017
 AlixPartners Industriegüterstudie 2018, URL: www.ots.at/presseaussendung/OTS_20180605_OTS0071/alixpartners-industriegueterstudie-2018, Accessed 20.11.2018
 Gartner: Top 10 Strategic Technology Trends for 2018, https://www.gartner.com/ smarterwithgartner/gartner-top-10-strategic-technology-trends-for-2018/, [20.11.2018]
 Shafto, M.; Conroy, M.; Doyle, R.; Glaessgen, E.; Kemp, C.; LeMoigne, J.; Wang, L.: Draft modeling, simulation, information technology & processing roadmap, Technology Area (11), 2010
 Grieve, M.: Virtually perfect: Driving Innovative and Lean Products through Product Lifecycle Management. Cocoa Beach, USA, 2011
 Siemens AG, Division Digital Factory: Advance Digital Enterprise – auf dem Weg zu Industrie 4.0, URL: www.siemens.com/content/dam/internet/siemens-com/customer-magazine/old-mam-assets/print-archiv/advance/adv152-de-screen.pdf, Accessed 20.11.2018
 Kuhn, T.: Digitaler Zwilling. Informatik Spektrum 40, 5, S. 440-444, 2017
 Datta, S.P.A.: Emergence of Digital Twins. Computing Research Repository (CoRR) abs/1610.06467, 2016
 Grösser, S.: Digitaler Zwilling, https://wirtschaftslexikon.gabler.de/definition/digitaler-zwilling-54371/version-189152, Accessed 20.11.2018
 Boschert, S.; Rosen, R.: Digital Twin ‐ The Simulation Aspect. In: Challenges and Solutions for Mechatronic Systems and their Designers, Springer, 2016
 Buchholz, B.; Ferdinand, J.-P.; Gieschen, J.-H.; Seidel, U.: Digitalisierung industrieller Wertschöpfung -Transformationsansätze für KMU: Eine Studie im Rahmen der Begleitforschung zum Technologieprogramm AUTONOMIK für Industrie 4.0 des Bundesministeriums für Wirtschaft und Energie, 2017
 VDI/VDE-GMA Fachausschuss 7.21., URL: i40.iosb.fraunhofer.de/
Digitaler%20Zwilling, Accessed 20.11.2018
 Petersen, K.; Feldt, R.; Mujtaba, S.; Mattson, M.: Systematic Mapping Studies in Software Engineering. In: Proceedings of the 12th International Conference on Evaluation and Assessment in Software Engineering, S. 68-77, 2008
 Ríos, J.; Hernández, J. C.; Oliva, M.; Mas, F.: Product Avatar as Digital Counterpart of a Physical Individual Product: Literature Review and Implications in an Aircraft, International Conference on Concurrent Engineering, S. 657-666, 2015
 OWL Maschinebau: Fachforum Digitalisierung konkret, https://www.owl-maschinenbau.de/asset/media/Einladungen/Einladung_Agenda_Forum_Digitalisierung%20konkret.pdf, Accessed 20.11.2018
 PLM Europe 2018, https://www.plm-europe.org/agendapresenterinfo/agenda.html, Accessed 20.11.2018
 it’s OWL, https://www.its-owl.de, Accessed 20.11.2018
Digital twin, Industry 4.0, asset administration shell, interoperability