How Digital Twins are utilised in industry
Digital twins enable completely new possibilities in different sectors such as industry or medicine. But what is a digital twin anyway? We discussed this question and what is already possible with such digital twins in the corporate and industrial sector in an interview with our experts Prof. Stefan Jack and Dr. Nikita Aigner.
What is behind the term “digital twin” and what is it good for?
A digital twin is not much more than a model. It can represent anything – cities, objects, machines, processes. In other words, a real object or phenomenon is described mathematically – when Sir Isaac Newton dropped an apple on his head, he developed the first digital twin of gravity. Today, these models are typically represented numerically in such a way that they can be calculated on a computer. So in essence, the digital twin is not new.
The essential difference to classical modelling – as with many modern technologies – is the fact that the computing power available today allows models to be calculated very quickly. Digital twins make it possible to represent reality in great detail and can also be coupled to real systems via sensors or other inputs. This allows us to better understand and analyse real systems.
How and in which areas can digital twins be used?
The list of useful application possibilities and areas of digital twins is very long.
In plant engineering, thanks to detailed modelling of machines – down to the level of individual drives – we can now make precise statements about their behaviour in a real environment. This allows us to plan entire production lines efficiently and to commission them partly virtually (i.e. from the comfort of our desks). This can save time, money and other valuable resources during physical commissioning.
During the operation of a plant, the feedback from sensor data alone provides simple visualisations about the status of the machines and running processes. Similarly, by measuring buildings, we can display the data in a 3D model. This gives us deep insights into the actual user behaviour and we can make adjustments in the context of renovation measures or better plan measures in terms of time.
But business processes can also benefit from digital twins. ERP (Enterprise Resource Planning) systems already digitally map large parts of companies today. And the systems are becoming increasingly powerful in terms of the fineness of the mapping. Thanks to this treasure trove of data, management can better overview and understand the complexity of their own company and further develop their own processes.
Digital twins are interesting when models from different domains are linked, e.g. production facilities and ERP. In this way, new relationships of data to each other and thus new insights or more accurate predictions are possible. If cycle times are automatically read out from the production process, the company knows precisely how expensive the production of a product was via the cost rates of employees and machines. This can be used in the next step to improve the preparation of quotations.
How can and will digital twins be combined with other technologies?
Since a digital twin is “merely” a model at its core, it thrives on synergies with other technologies. First and foremost is the computing power available today – on the one hand, we benefit from the massive performance available in the cloud, but even less powerful hardware can help us on the edge (at the edge of the data network) for the collection and pre-evaluation of data. But the value of new developments in communications technology should not be underestimated.
The coupling of a model to real-world inputs mentioned earlier can only be accomplished thanks to a variety of specialised communication protocols. On the one hand, transport protocols help us on the hardware side, which can be wired (fibre optics, Ethernet or fieldbuses for industrial networks) or wireless (from energy-saving LoRa to 5G high-performance networks). On the other hand, on the software side, communication standards such as OPC UA in the industrial environment, SiLA for laboratories and a variety of web-based protocols (JSON, XML, etc.) have significantly simplified implementation.
With the increasing complexity of digital twins, the results generated from them are also becoming increasingly complex. In the evaluation, we are thus increasingly dependent on methods from data science and machine learning in order to better understand correlations.
Are there also risks in using digital twins?
Of course there are. Digital twins can sometimes be very complex models that can interact with the real environment on top of that. They are typically tools for experts rather than end users. As for any modelling, it is always important to hit the right level of detail and make the right simplifications. When interpreting the findings from a Digital Twin, you also always need to be aware of what you are doing. Even if you are using a commercially purchased model of, for example, a plant, you still need to understand the limits of the system. Even the digital twin is not a magic black box.
About the people
Stefan Jack is Professor of Mechanical, Process and Manufacturing Engineering at the Digital Manufacturing competence area of BFH Technik & Informatik.
Dr. Nikita Aigner is a research associate at the Digital Manufacturing Competence Center and conducts research in the field of digitalisation with a focus on the woodworking industry at BFH Technik & Informatik.