![© Adobe Stock © Adobe Stock](/storage/media_foundry/images/4984/conversions/01-full.jpg)
Abstract
German industry is facing challenges: high wages and a shortage of skilled workers. The ‘Internet of Production’ (IoP) excellence cluster at RWTH Aachen University is developing solutions to overcome these problems and revolutionise industrial production. The aim is to achieve intelligent and efficient manufacturing by mapping real production processes in virtual form and networking all components.
Prof. Christian Brecher and his team are researching key technologies for ‘Industry 4.0’, including the use of machine learning and the creation of a ‘digital shadow’ that represents production processes in real time. These approaches are intended to lead to better quality, higher productivity and greater sustainability.
The IoP integrates more than 35 chairs and research institutions in an interdisciplinary manner and receives substantial funding. Despite the increasing digitalisation, the integration of people remains central. In addition, ethical issues and the protection of employee data are intensively discussed.
The research at the IoP could significantly increase the efficiency of German industry and provide an answer to rising wage costs and a shortage of skilled workers.
High wages, a lack of labour – how will Germany be able to maintain its position as an industrial centre in the face of these challenges? The Internet of Production (IoP) excellence cluster at RWTH Aachen University is working on solutions and thus paving the way for a new industrial era: real production is being mapped virtually, everything is networked with everything else and is becoming more intelligent. But people remain indispensable.
Alongside material and energy costs, wage costs are one of the key expenses of production. Unlike material and energy costs, wage costs in Germany do not fluctuate, but only go in one direction – upwards, as Frank Possel-Dölken notes. He is a member of the management board of the Lippe-based company Phoenix Contact, a manufacturer of solutions for industrial electrical and automation technology. He is responsible for digitisation at the company, which employs 22,000 people. ‘In the end, you have to ask yourself: are there ways to make these wage increases affordable?’ Although the company is active worldwide, it generates 70 per cent of its added value in Germany. According to digital head Possel-Dölken, technology must be used to find ways to maintain efficiency.
He is not alone in thinking along these lines. ‘There are large companies that say they have no chance of representing these wage increases,’ says University Professor Christian Brecher, Chair of Machine Tools at the Laboratory for Machine Tools and Production Engineering (WZL) at RWTH Aachen University. In 2022, wage costs in German industry were a good 44 per cent above the EU average. In addition, the problem of labour shortages will become even more acute when the baby boomers retire in the next few years. This will create challenges for science. ‘How can we make the conditions for industry in Germany even better?’ asks the scientist. ‘How can we involve people in this and try to be better than the others, step by step, so that wage costs are justified?’ The objectives are to achieve the best quality, higher productivity and more sustainability.
Network of top-level research
Brecher is the spokesperson for the IoP excellence cluster. In this network of top-level research, the mechanical engineering company works together with over 35 chairs and research institutions in Aachen on an interdisciplinary basis. They are paving the way for a new industrial future. The aim is to continuously develop a complex network of machines, software, data storage and people who can exchange data in real time and thus work intelligently. The IoP is one of 57 excellence clusters in Germany, funded by the federal and state governments with a total of 385 million euros per year. With the start of the second funding phase in 2026, the annual funding for up to 70 excellence clusters is to increase to a total of 539 million euros.
The Aachen research association is concerned with key technologies for the production of the future and thus with the core of the so-called ‘Industry 4.0’, which is often referred to as the fourth industrial revolution. After the introduction of the steam engine, electrical energy and the computer, the focus is now on a new dimension of digitisation: it should enable demand-oriented data analysis, e.g. through machine learning, and this should be applied holistically to all aspects of production technology: From product development, design and material selection, to manufacturing and assembly, and even new business models.
‘We have been looking at what people do for a long time. We have analysed what happens on the machine tool,’ says Christian Fimmers, managing director of the cluster, explaining the innovation: if the controller now finds a problem during a quality control, he can look at all the information from both machines and employees and check where the process deviates from the specifications. The data provides a virtual, real-time representation of the production process, a digital shadow of the real production. To achieve this, all the entities involved in the manufacturing process – people, workpieces and machines – are intelligently linked with each other using information and communication technologies. Ultimately, everyone should be able to communicate and cooperate directly with each other, including with the help of artificial intelligence (AI).
Making ‘big data’ more manageable
The researchers are doing the groundwork for this. Machines, sensors, machine control systems, but also humans themselves, are already constantly generating huge amounts of data that are referred to as ‘big data’. To make this ‘big data’ manageable, the scientists have developed an intelligent processing method: ‘We call this our digital shadow concept,’ says Brecher, illustrating the procedure using a projected tie. ‘In the shadow, we only see the outline and no longer the colour, but we still recognise that it is clearly a tie. This means that we slim down this digital world with the highest possible efficiency so that I can achieve my goal with as little data as possible.’ The researchers have now optimised this to such an extent that the calculations are incredibly fast, i.e. ‘in real time’. In addition, ‘look-ahead functions’ enable completely new ways of monitoring quality.
The importance of this in practice is illustrated by Possel-Dölken from Phoenix Contact: ‘The aim is to have very good information available at the push of a button – what is my current productivity, where is a problem occurring or developing?’ In the event of a problem, time is a crucial factor: it makes a difference whether it takes four weeks to find the causes or just one day. ‘In many cases, the difference is based on the availability of the right information. How long does it take me to access it? That is one of the key factors in increasing output in production and ensuring that good quality can be produced. This is a very central aspect, especially for Germany as a location.”
The researchers have developed a so-called conceptual reference infrastructure for the digital shadow, i.e. a framework for the design and implementation of digital shadows. This can be customised. The reference includes the development of models, processes and technologies for collecting, storing and processing data. One challenge here is the development of a common model for very different machines. ‘This is ambitious. But if we manage to do it, it would be a huge step forward,’ says Brecher. Two institutes are working on an exemplary milling machine and plastic injection moulding machine. Both have different controllers, processes and sensors. First, different models were created for each type of machine. ’At some point, you have to say: where are the similarities so that I don't have to keep re-inventing the wheel. What can be generalised, what can be transferred. Otherwise, the diversity of technology is unmanageable,’ says Christian Brecher, outlining the approach.
People continue to play a decisive role
And then there is the human factor. The Aachen researchers include this in the process from the outset – unlike in the 1990s when computer-aided manufacturing was introduced. “At that time, it was thought that everything could be automated and people were more or less ignored. In the end, that didn't work,” says Brecher. That is why the research association, which is otherwise rather technically oriented, now also includes the fields of psychology and occupational science, represented by Professor Verena Nitsch, among others. Nitsch makes it clear that, despite all the digitalisation, people continue to play a decisive role and describes the problems faced by a carmaker. The latter has automated and standardised an extremely large part of the production process. ‘Nevertheless, it was surprising to find that cars from different locations differ in the quality they produce. How can that be, when everything is standardised and automated? And it is exactly the same in both plants. So it couldn't be the automated part of the production process.’ The developments in IoP are therefore intended to support people, especially in their decisions within the production processes.
She comes to a sensitive point: if humans and machines are to be networked, if a virtual image of real production is to be created in real time, then data from humans is also needed. How far can and will we go in Germany? ‘China is of course more proactive in this regard,’ she says. Camera-based AI systems collect data there. ‘We are all right to ask ourselves: do we want that in Germany?’ says the ergonomics expert, addressing the area of tension: ‘We want more data and we need more data. When do we go too far? When do the disadvantages outweigh the advantages?’ Nobody should be monitored at every turn or have to fear negative consequences. But here too, the IoP's research activities are taking effect: reduced, anonymised data should enable the desired communication between companies. The researchers believe that this would solve many problems, for example within the supply chain. The IoP is working on this vision of communication between companies using reduced, anonymised data as part of a World Wide Labs project. This is also being done with regard to the symbiosis between humans and machines.
The aim is to identify requirements and problems and find solutions – as with employees who monitor automated processes and have to intervene quickly if anything goes wrong. The cluster has also been looking at this issue. ‘We looked at how we can use physiological data – for example, heart rate variability – to predict how alert a person is.’ If attention levels drop, errors may be harder to detect. This could lead to production stoppages, and in the worst case even accidents. Employee location data could also provide more information about when people need to be in certain places to ensure that production runs smoothly. Here, personality and co-determination rights, as well as ethical issues, play a decisive role.
Safeguarding valuable expertise
On the other hand, it is also about securing valuable know-how from experienced employees who are retiring. A classic example: a master craftsman takes a critical look at a component and decides whether it is good or bad. How did he do that, asks the younger colleague. The cluster is looking into how this individual expertise can be described and thus secured using a digital approach – possibly using data that is analysed by artificial intelligence and provided with features. Humans are very good at recognising patterns, explains the ergonomics expert Nitsch: ‘And if the machines are trained to recognise these patterns, then the machine can also learn to evaluate these decisions and suggest possible solutions for the decision-maker. And the next generation of employees can then learn from the machine again.’ But in the end, people would still have to make the decision.
Nitsch sees a number of positive effects: ‘They will then earn better money in better jobs, they will pay more taxes, which will benefit the country and the companies.’
In the second phase of the excellence strategy, which is currently being planned for 2026, the focus for the Aachen researchers in the ‘Internet of Sustainable Production’ will be on the aspect of sustainability in production technology in its various dimensions: economically efficient, socially acceptable and also ecologically sustainable. For example, the question of how long a product or machine has been used and how much potential it still has. ‘We drive a car for ten to twelve years. After that, it is of less interest to us, but it has the potential to live much longer,’ says Brecher, explaining what the focus will be. The prerequisite is a completely different product design: ‘It must be repairable and upgradeable, so that I can exchange some things and say: I can continue to use many components and still keep the product up to date, for example by using software updates. You can imagine that for many products.” On the other hand, the question of how long the components of a production machine will last has not yet been answered. “The data from a machine does not provide enough information. Statistically, I would need a huge number of drives to say that my component will last another thousand or 10,000 hours under these conditions,” says Brecher, outlining the problem. You would have to run thousands of hours of tests on a test stand to get a result. Unrealistic.
A question of risk and benefit
On the other hand, there are machines around the world that could provide this data – if they were networked. This would also help with completely different questions: for example, in comparing processes that lead to better results in one country than in another. We could learn from each other. This is where the Aachen researchers have to be persistent. Who would voluntarily disclose production data? For Brecher, it is a question of risk and benefit: ‘How do we convince companies that it is worth it for them to say: the added value for me in disclosing my data in a more anonymised form is worth more than the risk.’ The fact that the Aachen scientists go into companies with partial developments and demonstrate the benefits is certainly helpful.
In addition to a scientific advisory board, the cluster also has an industry advisory board with representatives from German and international companies, including BMW, MAN, Bosch, Siemens and Phoenix Contact. Furthermore, the cluster researchers are associated with industrial associations such as the German Machine Tool Builders' Association and the German Association of the Automotive Industry. At one of the annual meetings, Brecher was once asked when the Internet of Production could be seen. The answer makes the scale of this project clear: ‘It's too big a programme for a university and for us. We can't just play Google or Amazon,’ says Brecher, referring to the US cloud computing providers.
The focus is on basic research, impulses and transfer. The transfer of knowledge in the cluster is short-cut thanks to the connection to industry. And so the scientists offered entrepreneurs in this context: ‘We'll check how productively you use your machine.’ With the tools and methods they have developed in the cluster, they have identified significant potential for improvement - both with and without AI. This creates trust. Nevertheless, the new era of production will take time. Christian Brecher sees science as being ten to 20 years ahead of the times. Verena Nitsch emphasises that unexpected developments can also play a role – just as the home office, according to research, only became established in the 1980s with the coronavirus pandemic.
And the head of digitalisation at Phoenix Contact makes it clear how much stamina is needed to get to the Internet of Production. ‘It's a continuous process that takes decades to work on,’ explains Possel-Dölken. For major issues, it takes about five years before you see results and can reap the rewards. And then there is always the next step. ‘There is no state in which you can say that it is now complete.’