What is digital twin?

Turning “atoms into bytes” is a powerful way to enable rapid and responsive product development, especially when hardware is involved. Model based engineering is the foundation for this and a digital twin can be regarded as the next level up. A digital twin is a virtual model of a physical system capable of mirroring its static and dynamic characteristics. The virtual twin can represent a system that already exist in the real world or one that will be constructed in the future.

The idea of using “Twins” originates from NASA’s Apollo program, where at least two identical spacecrafts were built to reflect the conditions of the spacecraft during a mission in outer space.

The space vehicle on the ground was referred to as “Twin”. The Twin was extensively used for training during flight preparation. During the mission, the ground-based model was used to simulate alternatives, where the available flight data was used to reflect the flight conditions as accurately as possible and to assist astronauts in orbit in critical situations. The early idea of “Digital Twin” was introduced in 2002 by Michael Grieves as the conceptual model underlying future product lifecycle management (PLM).

You can make digital twins of any type of system: product, process or service. It can be the product you sell, the manufacturing plant producing it, the environment in which it is used, etc. A digital twin consists of various models of the physical counterpart of which some are executable (simulation models), and some are not (e.g. CAD models, electric schematics, configuration data, etc.), When connected to that physical system and collecting tons of data, either automatically or via manual feedback, it boosts innovation and performance improvement. Just have a look at the following short video.

Types of digital twin

One can distinguish 3 types of digital twin: DTP, DTI and DTA. Next to that, there is the Digital Twin Environment (DTE).

DTP – Digital Twin Prototype

This virtual twin is created even before the physical twin is there. It is therefore not linked to a particular existing physical asset. However, one could build proto test systems and them. It contains information necessary to describe and produce a physical version. It can be used to experiment, analyse, run automated tests, etc. especially when combined with virtual twins of the manufacturing process and usage environment.

DTI – Digital Twin Instance

To many a ‘true digital twin’. The virtual twin is linked to a specific physical asset. Operational data from that asset will be collected by the digital twin as well as its history throughout its lifecycle. This provides valuable information which can be used to predict future behavior and drive innovation and improvement. Also, the physical twin can be tuned improving performance and extend lifetime. A possible future state and behavior of the digital twin is sometimes referred to as a Predictive Twin.

DTA – Digital Twin Aggregate

This is a computing construct that has access to all DTI’s, i.e. an aggregate of a collection of DTIs. It can be used to make better predictions for failure and learn from general usage and variations.

DTE – Digital Twin Environment

This is an integrated, multi-domain physics application space for operating on Digital Twins. You could see it as the cockpit or central control room to interrogate and predict. Underlying is the digital thread weaving it all together. This is a mix of different technologies like PLM (Product Lifecycle Management) and IIoT (Industrial Internet of Things). It connects physical twins (preferably smart devices for automated interaction), digital twins, people and processes.

Building your Digital Twin Environment

The main ingredients for a digital twin environment are models and a framework to manage and interact it all. Many organizations already have invested in all kinds of models and will continue to do so. Also, many organizations have invested in PLM systems which serve as a framework. The speed at which PLM and modelling tools develop, is incredible. For one, you see that PLM is becoming cloud based (SaaS) allowing you to work and connect from anywhere, and more human centered making it easier to use and manage. At the same time modelling becomes much more easier and affordable. Just think of generating augmented instructions with the push of a button or CFD computing as a service where you just pay by the hour for the speed you need.

Where to invest and how far to go depends on the business case behind it which fortunately is rapidly becoming more attractive. Like building your product, you build your digital twin incrementally and iteratively based on value for effort (‘bang for buck’). A good place to start is identifying what’s already out there regarding models, data, supporting systems, digitalization knowledge, etc. Next would be to find some low hanging fruit to demonstrate benefits and experience what is needed to make it work.