Its Keen Eye Doesn’t Miss a Thing!
It’s clear that artificial intelligence is extremely powerful. There’s no doubt
about that. But can it also see into the future to prevent problems? Yes, say the
experts from Digital Production at LANXESS. They are currently working on
the launch of a new tool designed to prevent unplanned downtime and the
resulting damage to our systems.
Mr. Recker, artificial intelligence is currently being touted as a panacea for every problem. What opportunities do you think the use of AI will bring for our production lines at LANXESS?
Sebastian Recker: We do indeed see a great deal of potential here. We’ve been exploring the possibilities of AI-supported tools for quite some time and are currently conducting pilot projects at several plants – the results so far have been quite promising. In fact, we’re now already moving ahead with the implementation of the first measures – in the next three years, we want to be able to monitor the 1,000 most critical plant components at LANXESS with the help of AI. This includes vital components and machines, such as our exhaust gas incinerators, whose failure would result in a production standstill. Through what is known as predictive maintenance, we will be able to minimize unplanned downtime and the associated production losses, high costs, and damage to our systems.
Can you explain how this will work in more detail?
Bastian Jana: We are training the AI with the values from our OSI PI system, the central archive for all of our production data. As a result, it learns what the sensor data normally looks like when the production line is running well. It then monitors all of the parameters in real time and sounds the alarm as soon as it detects deviations. This may indicate a component that is developing a defect. Together with the employees at the plant, we can then take a close look at the relevant area of the system – and verify whether our suspicions are confirmed. We are currently in the process of setting up the models for the first plant sections. To this end, we have set up a project team under the direction of PTSE Central Maintenance. For the time being, the project involves a lot of extra work. But we’re absolutely sure that the effort will pay off.
Are there other possible applications for AI in production besides maintenance?
Recker: Yes, by using it not for the equipment, but to improve the process. The aim here may be, for example, to use fewer raw materials, achieve a higher degree of purity in the finished product, or reduce energy consumption. The AI can use historical data from the production process to determine the best settings for a given situation. We’ve already been very successful with this in Kallo, and have identified potential savings. The field of predictive analytics is still in its infancy, however. We have to carefully consider what additional value a tool like this would offer us. Because ultimately, more data also means we need more experts who can evaluate and analyze it.
Sounds exciting. But are there limits to what AI can do?
Jana: Yes, AI has its limitations, especially in situations with many changing products and where new processes are constantly being introduced, as is the case, for example, in the business of our Saltigo BU. In the case of standardized processes that always remain the same, AI can certainly be a driver in maintaining a consistently high level of quality. What it can’t do, however, is to set up a new process from scratch. This still requires our operational experts’ know-how.
Recker: Thanks to the data-driven approach that we at LANXESS have paved the way for through extensive digitalization in recent years, we are definitely in an ideal position to use AI. There are sure to be many more advances in this field in the years to come. That said, we will be carefully scrutinizing every step we take. This is because our aim is to support our employees in their work with our tools, not overwhelm them with ever-new dashboards. The work of our Digital Working Group Production, an interdisciplinary team that manages digital activities in production at LANXESS, also plays an important role in this regard.
Digital Production at LANXESS
- LANXESS has been hard at work digitalizing its production systems since 2016. All the tools defined for the roll-out, such as OSI Pi as a central archive for all production data, TrendMiner for data analysis, and FINITO as a digital shift log, are now available at all plants worldwide.
- In addition, other tools are currently being rolled out, such as COMOS for digital system documentation or Mobil Worker for digital checklists in production and maintenance.
- The OSI PI system archives the readings from more than 600,000 sensors in LANXESS production lines. More than 1,000 employees worldwide are already using LANXESS’s TrendMiner server.
- For more information, visit the Xnet page entitled “Digital Production.”