A Digital Twin is a virtual representation of a physical system, like a CAD model for a machine. But instead of it just being a model for, it is also a model of. When you look at a virtual representation, you know exactly what the current state of the physical system is and, to take it a step further, use the virtual representation to test, change and improve its real-life counterpart. Digital Twins go beyond being mere models or simulations as their essence is that they are complete representations of physical systems making them fully interchangeable.
Or as it is described in an excellent article (see the links at the end of this article) by NASA:
“A Digital Twin is an integrated multiphysics, multiscale, probabilistic simulation of an as-built vehicle or system that uses the best available physical models, sensor updates, fleet history, etc., to mirror the life of its corresponding flying twin.” (…) “By combining all of this information, the Digital Twin continuously forecasts the health of the vehicle or system, the remaining useful life and the probability of mission success. The Digital Twin can also predict system response to safety-critical events and uncover previously unknown issues before they become critical by comparing predicted and actual responses. Finally, the systems on board the Digital Twin are capable of mitigating damage or degradation by activating self-healing mechanisms or by recommending changes in mission profile to decrease loadings thereby increasing both the lifespan and the probability of mission success.”
NASA and the USAF are probably the first implementors of the Digital Twin concept which should give you a pretty good idea of the kind of environment Digital Twins are best suited for; complex, high-risk, high-cost physical systems that have a long lifespan. On the other side of the spectrum is the use of Digital twins in regular manufacturing. Its main proponent is Michael H. Grieves who has introduced the name Digital Twin in 2003 at his University of Michigan Executive Course on Product Lifecycle Management (PLM). In an excellent paper, Digital Twin: Manufacturing Excellence through Virtual Factory Replication he states that the conceptual model exists of three parts a) physical products in Real Space, b) virtual products in Virtual Space, and c) the connections of data and information that ties the virtual and real products together. Griever’s view on the use of Digital Twin is business oriented:
“Focusing on the connection between the physical product and the virtual product enables us to conceptualize, compare, and collaborate. We can conceptualize visually the actual manufacturing processes. We can compare the formation of the physical product to the virtual product in order to ensure that what we are producing is what we wanted to produce. Finally, we can collaborate with others in our organization and even throughout the supply chain to have up-to-the-minute knowledge of the products that we are producing.”
According to Gartner Digital Twins, although not new, will be a big thing in the coming years as “Several factors have now converged to bring the concept of the digital twin to the forefront as a disruptive trend that will have increasingly broad and deep impact over the next five years and beyond.” Returning to the conceptual model of Grieves we can easily see what Gartner means by ‘factors that have converged’:
(a) The physical products in Real Space are becoming more and more complex and distributed, especially in large manufacturing industries. This makes new or improved ways to make them more manageable a necessity.
(b) Virtual products in Virtual Space have been made possible by the introduction of the computer and related hard- and software. But to create true-life virtual products requires very high-performance hardware and software systems, which have greatly improved over the past years.
(c) Recent developments in the connections of data and information that ties the virtual and real products together are the biggest contributor to the rise of Digital Twins. Only with the availability of Big Data, IoT, Machine Learning and related technologies this tying of virtual and real products can be done to such a degree that Digital Twins can become a reality.
If you combine the Nasa and USAF definition of a Digital Twin, the technology they work with and the kind of budgets they have available it is clear Digital Twins don’t come easy nor cheap. To build true digital twins requires a huge investment (in terms of expertise, time and money) only a few can afford. But as Gartner research vice president Alfonso Velosa explains; for businesses “The complexity of digital twins will vary based on the use case, the vertical industry and the business objective.” And regardless of complexity ““(…), to truly drive value from digital twins, CIOs will need work with business leaders to develop economic and business models that consider the benefits in light of the development costs, as well as ongoing digital twin maintenance requirements.”
Deloitte has written an insightful article on the applications of Digital Twins in manufacturing (see the links at the end of this article) giving a general oversight of the concept and how to get started. The figure below offers a clear visualisation of the relationship between Digital Twins and their physical counterparts. The relationship is a vicious circle where physical systems communicate with their Digital Twin, providing (raw) data which is then aggregated and analysed. The resulting (digital) insight is used to modify the physical system if necessary.
On a more technical level, the Digital Twin architecture (see the figure below) is informative of the data an technology involved. The first thing to note is the importance of Contextual Information; physical systems are used in contexts which influences their design, application and life cycle (for example; a windmill’s contextual information may include not only relevant weather statistics, but also relevant location data like closeness to houses). Another thing to note is the importance of Artificial Intelligence needed to make sense of the huge amounts of data in a fast and reliable way.
By now it should be clear that designing and using Digital Twins is a daunting task not to be taken lightly. Implementing them requires specific knowledge on both the business domain of physical manufacturing and the various technology domains in the digital realm. Right now only companies like IBM, GE Digital and Siemens are able to deliver the kind of expertise needed. But as this video by Siemens of a real-life application of Digital Twin technology shows the results can be impressive:
The NASA/USAF article on Digital Twins:
Michael Grieves webcast and article:
Deloitte’s excellent oversight: