Chat over coffee - Prof. Dr.-Ing. Otto Appel meets Jochen Hirdina and Robert Aust at Krones AG in Neutraubling

What does it mean for a machine and plant manufacturer to enter the field of AI? Is it no longer an alien world for mechanical engineers?

Hirdina: Today we can say quite clearly: no, it’s not. Krones has been working with software solutions for the beverage industry for several years. However, it was clear to us early on that there were tasks that could not be implemented with classic software solutions, or only inadequately, and that an AI approach should be tested here. Interdisciplinary teams were quickly found and together they developed solutions.

 

What preparations were necessary, in terms of time and personnel, to take Contiloop AI from development to a first field test? I am thinking - for example - of the immense effort of the VW subsidiary Cariad to establish AI in the automotive group with about 6.500 people.

Aust: Indeed, something like this cannot be developed in a hurry. We already started dealing with the topic of AI in 2017. AI experts had to be found and trained who not only understood the pure AI application, but could also delve deeply into the machine technology together with colleagues from the product areas. Behind the resulting Contiloop AI application is much more than just a piece of software that independently controls certain processes of the stretch blow moulder intelligently. Rather, there is a whole infrastructure behind it that has to be built and maintained so that this solution can be scaled and made available to many of our customers.

 

AI doesn’t seem to have yet found its way into the PET industry. Is this the case and does Krones see itself in a pioneering position?

Hirdina: We had a clear use case in mind that we wanted to solve with an AI approach. When the first promising successes became apparent during development, this spurred the development team on even more to bring the product into a field test as soon as possible. And so it came about that a functioning system was already in productive use at a customer since the beginning of 2022 and we were able to present Contiloop AI at drinktec in September 2022.

 

What problems can customers eliminate or reduce by using Contiloop AI? What are the advantages for the customer?

Aust: With Contiloop AI, we are addressing several of our customers’ challenges at the same time. Mainly, we want to relieve the machine operator of one of the most demanding tasks he has - maintaining bottle quality during production. This means that the system monitors inline the material distribution of each bottle produced and - if necessary - adjusts the stretch blow-moulding parameters fully autonomously, so that the machine operator can concentrate fully on other activities. Some of the advantages that result are:

• Operators do not have to make manual adjustments when environmental conditions change.
• Offline quality checks on bottles can be reduced to a minimum.
• Fewer highly skilled process specialists are needed.
• Less variation in material distribution of bottles produced.
• This results in a much lower rejection rate of poor quality bottles.
• Very stable stretch blow moulding process and thus greater potential for bottle weight reduction and improved processability of recycled PET material.

 

As a customer, do you need to understand AI to reap its benefits?

Hirdina: No, an essential criterion in the development of the system was also to make the inner complexity of the system nevertheless very simple to use. The best AI solution is one that you don’t even actively notice exists.

 

Contiloop AI is a cloud-based application from the field of artificial intelligence. Don’t your customers’ IT departments have problems with this? What certifications are required for an AI system?

Hirdina: Of course, as with any new technology, there are certain reservations on the part of some customers. That’s why we are completely transparent and explain to our customers exactly what data and information about the machine and the process are required and stored in the Krones IIOT platform in order to be able to operate the system to maximum advantage. Furthermore, we cooperate with the German Federal Office for Information Security and PricewaterhouseCoopers to ensure the necessary trustworthiness (“Trustworthy AI”) of the system.

 

What is the long-term corporate strategy and what are the advantages for Krones?

Aust: In the long term, we want to use AI solutions where they offer advantages both for our customers and for Krones itself. By using it in the production environment and the information gained from it, we can further develop our machines faster and in a more targeted manner, or support our customers more precisely in solving problems.

 

Contiloop AI in the test field: what can you report about experiences, successes and the first steps?

Aust: The experience we have gained so far in the production environment shows that we have already been able to overcome some of the challenges. However, we also see that there is still potential for optimisation, which we will improve by continuously developing the system.

 

The complexity of producing a PET bottle is difficult to grasp. If AI takes over the job of bottle production and quality, then - to put it simply - operator know-how must be made manageable via sensors and sensor technology. How does that work?

Hirdina: We are currently focusing on the most important parameters with the system, which a human operator would also change if he had to adjust the process. The system is already superior to a human operator in terms of accuracy and speed. In the future, we will integrate further existing parameters into the AI logic, but also further optimise the system with new sensor technology that is not yet available, if this should be necessary.

 

In the long term, does it mean that a stretch blow moulding machine or a filling line without AI will no longer work? I also ask this question from the point of view that you, as a machine supplier, make performance commitments for output, scrap rate, energy consumption, bottle quality, etc. based on specifications.

Hirdina: It is not Krones’ declared goal that AI solutions will be absolutely necessary for the operation of filling lines. However, in the long term it will be the case that with their use a significantly higher level of effectiveness will be achieved than before, and performance commitments could presumably be linked to the use of such systems.

 

Can third-party machines and systems be included?

Aust: If integrated third-party machines have the necessary interfaces, we will consider them as part of the overall system and integrate them accordingly if it makes sense to do so.

_______________

''The experience we have gained so far in the production environment shows that we have already been able to overcome some of the challenges.'' - Robert Aust

_______________

Isn’t it technically fascinating that in our industry today’s block concepts are still being questioned as an efficient mechanical way to combine machines, while on the other hand the potential of AI in these processes is already being discussed?

Hirdina: Indeed, state-of-the-art technology is not equally advanced in all regions of our planet. But Krones wouldn’t be where we are today if we didn’t actively drive innovation and thus set benchmarks in our industry time and again, in keeping with our motto - Solutions beyond tomorrow.

 

Thank you for the spontaneous meeting and the coffee.

www.krones.com