Q & A with Dr.-Ing. Volker Trinks, SCHOTT AG
Why are assistance systems an important topic for SCHOTT?
We are the world market leader for special glass tubes for pharmaceutical primary packaging materials such as glass containers for vaccines. In this area, quality requirements have been increasing for years. However, ensuring consistent quality requires a great deal of effort in the case of tubular glass, because every draw, meaning every manufacturing unit, is individual. In order to optimally adjust processes, which are also very individual, and to continuously monitor the result, we use optical technologies and intelligent methods of image processing and pattern recognition. Without such methods, the individuality of draws can hardly be captured.
Why is intelligent assistance indispensable in these areas?
There are several reasons. Setting up production processes in the best possible way has become extremely complex. Many parameters play into melting and shaping. Only highly qualified specialists are still able to model optimal interaction. Operators on the lines, on the other hand, need assistance systems that determine the best parameters possible at the specific moment to run a process. Such systems must also signal whether there is still potential for optimization. As far as quality assurance is concerned: During production, defects in the glass must be continuously detected and clearly classified in order to guarantee product properties with 100 percent accuracy. This is not possible with the naked eye, but it is possible with optical technologies combined with AI. With at least semi-autonomous control and optical inspection, we therefore achieve a very high level of quality across all sites. In addition, we secure the know-how of process experts to a certain extent.
Won’t this make people on the line dispensable?
Definitely not in the near future. Digitization will not replace a single employee in production at our company, because only people can react flexibly to unusual situations. An important task is therefore to design increasingly complex systems in such a way that machine operators can use them well and receive all the information they need. For example, it must become transparent why a system makes a certain decision.