Digital twins for the machine industry

Learning digitally is as good as learning on a real machine

Extensive training is required to operate the complex machines of grinding machine manufacturer Reishauer. In a test, 20 people were trained directly on the actual machine and a further 20 were trained digitally using a digital twin in virtual reality.

Using the conventional method, the machine is not available for production during the training. The operators were taught in groups of three which took 2.5 hours each.

In contrast, the digital-twin-based method could be carried out independently by all operators at any time and place. The operators only needed 50 minutes to do this. Interaction with the digital twin was intuitive through hand and pointing movements, which were captured by sensors from Sensoryx.

One week after the digital-twin-based or conventional training, the operators were tested on the real machine. It turned out that the tasks were solved equally well by both groups. In summary, digital training has the upper hand: it is faster, can be carried out independently of time and location, and the real machine can produce without interruption. 

Good simulation of potential collisions

Grinding machines for gears must work with micrometre accuracy, which requires precise part guidance. Complex algorithms are developed to prevent a collision between the part to be ground and the machine during the grinding process. The second study aimed to determine whether such collision analyses could be calculated more easily, faster and with less effort using digital twins. Here too, the results are very satisfactory because the required accuracy in collision detection was achieved, and the analysis was performed at the push of a button, saving time.

Technological challenge

What sounds so simple was a significant technological challenge: a single digital twin had to be created from the three individual components machine, controller, and algorithm. In other words, a digital version of the Reishauer machine had to be built, equipped with the Siemens controller and the complex algorithm.

Everyone benefited

It took all the partners involved to realise this project. The ETH researchers led by Professor Andreas Kunz were able to further develop and apply their VR research in practice. Alexander Farger from Siemens concludes: "Machine manufacturers and operators are already successfully using our products 'Create and Run myVirtual Machine' to create digital twins and optimise manufacturing processes. However, we also see great potential in using digital twins in the training of operators and maintenance personnel. The combination with virtual reality takes the teaching of training content to the next level. The solution is also interesting for educational institutions as it provides an easy way to learn how to use machines of different technologies and complexity levels.

Florian Hänni from Reishauer sees several advantages of this project: "Thanks to the support of Innosuisse, we were able to evaluate two options for collision testing. Both worked. We will integrate collision testing into our design software but will use a different technological approach. Furthermore, we have learned that it is possible to convey complex tasks virtually with a digital twin." For both Hänni and Farger, projects with ETH Zurich are of high value because they knew from the beginning that they had access to the experts in this field and that communication was always open and on an equal footing.