How far should we go in connecting processes and smart technology and what would the return be?
For me, the promise of Industry 4.0 and the smart factory of the future is that you connect previously unconnected silos of data, which gives better visibility and connectivity of data. The intuitive answer to this question would be to connect the two as far as possible. What I also learned is that, very often, when you apply more advanced machine learning algorithms to analyze data, often the surprising effects come from the fact that you find connections between data that were not connected before and where the algorithm finds patterns that were not visible to the human expert. However, we can also take it too far. There can be connectivity just for the sake of connectivity, chicken and egg problem. We don't know some of the surprising patterns we would find as a result of connecting data. The approach here should be one of experimentation. We start by connecting some of them, really looking at the results in these use cases and then adding more and more.
How can we deal with more product and process complexity? How can we enhance flexibility?
Very often the question is how we can use these connectivity insights that have been generated? How can we use it to achieve certain business objectives? Very often, when I talk to the executives at my Leading the Factory of the Future Masterclass, initial intuition of managers is that they do it for operational efficiency with regards to reducing costs, having predictive maintenance, lower unplanned downtime and so on. However, I always encourage people to really think beyond operational efficiency, and address strategic questions, like flexibility. An interesting story to share here at this point, we were guests of BMW, one of the most advanced BMW plants in Leipzig, as part of the Masterclass we’ve now been visiting for over three years. It's always really interesting to see how they develop and for them one of the key topic is sharing five core objectives why they do industry 4.0 at this plant and one of those was cost reduction and they were achieving around five per cent cost reduction. But the majority of the thinking was to enhance the resilience of the plans to have flexibility with regards to volume and the product mix. In the past, for BMW the strategy was that we have a global network of plants and you share the series between the plants. One plant is dedicated to the x-series, another for the SUV's and so on. It was easy to optimise in this setup, however, giving the current political situation, where you have a Brexit coming but no one really knows when, and you have an American president with new kind of tariffs and trade barriers, our main objective is now to have a much higher volume mix flexibility. Also, with regard to electrification of the cars, everyone is sure in the industry that electrification will come but no one really knows at what point there is a tipping point when consumers will start purchasing them, which is really challenging for volume flexibility.
For an ambitious industry 4.0 strategy, you should aim for flexibility.
What about the manager role? So how will the Industry 4.0 change the job of a manager in operations?
What's the job of a manager? To make decisions. Leaders I meet during Leading the Factory of the Future Masterclass say that AI algorithms can never replace what they do as it requires their domain expertise to really lead effectively and make critical decisions. There is another camp in academia
believing that maybe algorithms do a better job as a manager in making operative decisions as the algorithm is not biased by perceptions of the world, has no past dependencies, Monday mornings or Friday afternoon feelings. My vision is that we will have a mix. We will probably use advanced machine learning to provide the manager with a clue, a hint, an indication that there is an interesting pattern and then we enable a manager a tool allowing for experimental learning on a digital twin in a very advanced simulation model. So we would use that at this point. We would amuse artificial intelligence to find interesting indication interesting clues and then use human intelligence to come up with a solution proposal or some ideas on how to implement it and then again, we would use the simulation system and an AI to really simulate these ideas of the manager and find out about the consequences and do iterations. I would say the role of the manager, in the end, is exactly the same.
However, you would start probably on a higher level and somehow co-create your decisions in a team with an algorithm. Until we are in this ideal stage, I would say the main job of the manager today for industry 4.0 is to really initiate and lead the change process to inspire people to create the vision with the team.
What about the actual solutions being talked about, are there solutions in the field of digitization not worth trying e.g. AR in commissioning?
There are a lot of hyped technologies. I know companies that are very happy about the augmented reality and its applications, and many have tried it and said it’s still very complicated, it’s very error-prone and actually is distracting the operator more than helping them. I would say there are no technologies that are not worthwhile trying. I think the important thing is to come up with your use case and my recommendation here is that use case should never start with technology. It should really start with an open problem. It should start with a problem, either of the operator, the worker, an engineer and then we should think why can't we address this problem with current technologies, with the way we use it today? And then we may look into the huge assortment of novel, digital tools for decision support, for optimization, and then perhaps explore some of the new tools. I think the mistake that many companies have made is that they were fascinated by just the technology like augmented reality, then just invest in the technology and use lots of good use cases and then we say the technology is not good, while actually it was our question that was not the right one. So my advice here is to always start with the questions, the problems and then see in an experimental way how technology can help.
How can you drive mindset change in brownfield sites, which are performing well?
This is one question we discuss often in our Masterclass. The idea is that, we know about new opportunities, we read the industry reports, we know that we should do something. On the other hand, we are working with full capacity, we are making good money, we still a list of conventional established projects ahead of us for continuous improvement. We have a stable state, why should we really disturb all that and look into the future? How can we motivate ourselves to look into this and it's a challenging question, not so much for top leadership, but more for middle managers or the people that are then really in charge of the implementation. There are different ways to address it. I think first of all we have to really see that it's the best situation if you have well-working brownfield as it's always much better to manage and pilot something and explore something new out of the situation of strengths and not out of a situation of life.
People always tend to act following a crisis. We should be more proactive and not just reactive. If I have a good operating brownfield, I may have a strong lean practice and actually, a strong lean practice is very supportive for digitalization as it makes no sense at all to digitalize a bad process. If I have a brownfield that has been optimized and has good practices, then it's a perfect situation to think about how can we reach the next level and optimise. Beyond single workstations, looking into the gains of connecting the data of different applications and there are different ways how to do it, you don’t want to interfere with the entire well-optimised process line. But you will usually have pilot plant, special lines for special orders and customers requiring a higher degree of customisation. There are centres where we explore one or two new technologies, perhaps we just invested in a new piece of machinery, which has better connectivity capabilities. Often it makes sense to really pick these pockets, which are a little bit different from the rest of the organization and start your operations there. I have another concrete example on that, again from the BMW in automotive manufacturing. They started to explore cobots. They have a very well optimized chain and the assembly has a clear tact structure. They started with a cobot it the parallel station, where you do repairs, to apply it in emergencies, instead of starting with the main assembly line. In the normal process, you place a car in a special workstation to perform some manual work before putting it back on the line. Cobot was used to apply an adhesive, which is a job that requires high precision. A year later, the cobot was on the assembly line, used as part of the standard procedure. These are the ways you can also optimise brownfield processes, starting exploring new digital technologies.
We started to use an experiment with Robotic Process Automation, any recommendation when to use it via solving a problem versus a standard IT project?
Using an example I just shared from BMW, in most industries we have well-established practices where we start utilising robots in strict robotic environments. Most plants I’ve seen across the industries have started experimenting with utilising robots, cobots specifically. For example, in a Siemens factory I visited, they did assembly of more complex electric motors and they didn’t have any cobots yet but they were interested in the technology. One of the leaders gave a cobot as a Christmas present to his team, asking them to play around with it using iPads to programme it. One and a half years later, as a result of placing the cobot as part of one of the six sigma cycles, this cobot was in the line. It’s a great example a bottom-up approach to introducing new digital technology.