Raphael Thys
Fig 1: The Mirrored Spaces Model by Grieves [1]. [Image Source: Leon Eversberg]
Introduction
The term Digital Twin has gained a steady increase in interest over the last decade according to Google Trends, see Fig. 2.
Fig. 2: Google Trends plot for the search term “Digital Twin” over the last 10 years. [Image Source: Leon Eversberg]
One of the many reasons is the growth of key enabling technologies, such as Artificial Intelligence, the Internet of Things (IoT), and the Industrial Internet of Things (IIoT) [2].
But, if you ask three people what the Digital Twin is, you will get four different answers. So, what is a Digital Twin?
In this article — which is the first part of a series — I cover the history of the Digital Twin term, its definition, and what has to change for the Digital Twin to become widely adopted.
What Is A Digital Twin?
The concept of the Digital Twin goes back to a presentation about Product Lifecycle Management from Dr. Michael Grieves in 2002 at the University of Michigan.
As shown in Fig 1. at the top of this article, the original concept consisted of a physical and a virtual space. The virtual space uses data from the physical space as input. And the physical space uses information from the virtual space. Thus, there is a bidirectional connection between physical space and virtual space.
John Vickers coined the term “Digital Twin” later in 2010 and then NASA adopted it in a technology roadmap. Contrary to what some sources say, Digital Twins were not used in NASA’s Apollo program [1, 3].
Digital Twin Definition
The Digital Twin Consortium defines a Digital Twin as follows [4]:
“A digital twin is a virtual representation of real-world entities and processes, synchronized at a specified frequency and fidelity.”
Looking at the scientific literature, I prefer the often-cited definition of Kritzinger et al. [5]. Fig. 3 defines the difference between a Digital Twin, a Digital Model, and a Digital Shadow.
Fig. 3: The Digital Twin according to Kritzinger et al. [5], [Image Source: Leon Eversberg]
Fig. 3 is an extension of the original concept of Grieves. A physical object and a digital object are bidirectionally linked to each other. However, the data flow can be either manual or automatic.
In a Digital Model (e.g., a CAD model), information flow is always manual.
A Digital Shadow automatically receives data from the physical object. For example, using sensors for real-world measurements. But, the physical object is not automatically affected.
The Digital Twin has automatic data flow in both directions. This means that the Digital Twin will usually have some kind of simulation or AI-based prediction outcome to provide feedback to the physical object.
Because the Digital Twin originates from Product Lifecycle Management, it is supposed to handle a wide range of its lifecycle data from cradle to grave. For example, data from engineering, production, and customer service.
Why Use Digital Twins?
Industry 4.0 is characterized by cyber-physical systems in which software systems, products, and machines communicate directly with each other. The Digital Twin technology is expected to be a key enabler of these Smart Factories because it provides interoperability.
In today’s world, we all know how valuable data is. In practice, however, needed data is often stored in different places and with different data formats. This leads to so-called data silos, where information between two systems can only be exchanged manually. The Digital Twin promises a unified Application Programming Interface (API), connecting different data silos and software systems [6].
Is the Digital Twin Just a Buzzword?
Gartner’s annually released Hype Cycle for Emerging Technologies, depicted in Fig. 4, shows the common progression for new technologies and innovation. According to Gartner, the Digital Twin technology was at the Peak of Inflated Expectations in 2018.
Fig. 4: Gartner’s Hype Cycle for Emerging Technologies [7], [Image Source: Leon Eversberg]
At this peak, expectations are much higher than the technology's actual capabilities at that time. Thus, the Digital Twin could be described as over-hyped back in 2018.
At the same time, Gartner also predicted that the Digital Twin will be at the Plateau of Productivity between 2023 and 2028. At this plateau, the technology’s benefits have been demonstrated in real-world use cases and more and more companies start to adopt the technology [7, 8].
I think the Digital Twin is currently somewhere on the Slope of Enlightenment.
Why Are Digital Twins Not Widely Used Yet?
A key challenge for the adoption of Digital Twins is the current lack of standardization [2]. Standardization is key for Digital Twins to establish communication with different entities, as well as semantically understand its data.
To this end, the ISO 23247 Digital twin framework for manufacturing has been published last year in 2021. The four parts cover general principles, a reference architecture, digital representation, and information exchange. A more specific standard for Digital Twins in Industry 4.0 is still under development as IEC 63278: Asset Administration Shell for industrial applications.
However, establishing open standards for Digital Twins is not enough. We also need software solutions.
There are multiple international organizations currently driving Digital Twin adoption by providing open specifications and open-source code [9]. The Open Manufacturing Platform, the Digital Twin Consortium, and the Industrial Digital Twin Association all have GitHub accounts with code. For example, the GitHub page admin-shell-io by IDTA provides open-source tools to create and serve Asset Administration Shells.
Conclusion
A Digital Twin is a virtual entity, which is bidirectionally linked to its physical counterpart.
Now that Gartner’s “Peak of Inflated Expectations” is overcome, we can expect the Digital Twin to slowly reach the “Plateau of Productivity” in the coming years.
According to a market analysis by Research Nester, the global Digital Twin market is estimated to grow from 2022 to 2031 with a compound annual growth rate (CAGR) of 44 % [10].
Standards and open-source software are starting to enable the adoption of the Digital Twin. However, the technology is still far from widespread usage. I predict that this will change over the course of the next few years.
The next article in this series will look at open source software for actually using the digital twin.
How to Use the Asset Administration Shell
A tutorial on creating and interacting with digital twins using AASX Package Explorer and AASX Server
References
[1] M. Grieves, Physical Twins, Digital Twins, and the Apollo Myth (2022)
[2] A. Fuller et al., Digital Twin: Enabling Technologies, Challenges and Open Research (2020), IEEE Access, vol. 8, pp. 108952–108971
[3] M. Grieves and J. Vickers, Origins of the Digital Twin (2016)
[4] Digital Twin Consortium, Definition Of A Digital Twin, accessed: 30.12.2022
[5] W. Kritzinger et al., Digital Twin in manufacturing: A categorical literature review and classification (2018), IFAC-PapersOnLine, vol. 51, Art. no. 11
[6] S. Malakuti, The digital twin: from hype to reality (2021), ABB Review, accessed: 30.12.2022
[7] Gartner, Gartner Identifies Five Emerging Technology Trends That Will Blur the Lines Between Human and Machine (2018), accessed: 29.12.2022
[8] Gartner Research, Understanding Gartner’s Hype Cycles (2018), accessed: 29.12.2022
[9] E. Barnstedt et al., Open Source Drives Digital Twin Adoption (2021), Industrial Internet Consortium’s Journal of Innovation, pp. 19–34
[10] Research Nester, Digital Twin Market Analysis (2022), accessed: 30.12.2022