Research project: using AI to create resilient businesses

Disruptions in production and the resulting quality losses in products occur repeatedly. Machine breakdowns, technical malfunctions, changes in material composition, leaks, escaping lubricants, etc., the causes are diverse and often unpredictable. The better and faster a company can respond to these influences, the more resilient, that is, robust and ultimately more competitive it becomes. Within the SPAICER research project, Smart Resilience Services (SRS) are being developed using AI methods. They are designed to enable companies to detect disruptions, including in production, more quickly and respond more effectively. Ideally, it will be possible to avoid these disruptions altogether, or at least reduce them.

The SPAICER project is funded by the Federal Ministry for Economic Affairs and Climate Action. The consortium consists of researchers, industry partners, and supporting IT and software service providers. Coordination of the research project is carried out by the German Research Center for Artificial Intelligence (DFKI) in Saarbrücken.

Wear forecasting - determining the ideal maintenance time

A key factor in recognising wear is the condition monitoring of production machines and predictive maintenance based on real-time data collection.

Within SPAICER, one use case focuses on determining the optimal point at which a tool is worn enough to require replacement, but still functional enough to maintain consistent product quality and a smooth production process. This use case is led by Feintool Systems Parts , which manufactures components for industries such as automotive using fine-blanking processes.

To obtain detailed data on the fine-blanking process for the research project, vibro-acoustic diagnostics are used, an innovative acoustic method. AI analyses these data to create wear forecasts, enabling staff to be informed in advance about upcoming maintenance, for example the replacement of “active elements”.

The benefits are clear. Automated wear forecasting makes maintenance and production processes more predictable. This saves time and costs associated with regular tool inspections while also improving service quality as a supplier and manufacturer.

Optimising machine parameters

Another key aspect is adjusting the straightening machines and fine-blanking tools. Whenever new steel coils are delivered, the machines and tools must be set according to the microstructure of the steel strips. This consumes time and material, as up to 100 or more parts may need to be produced and checked before the desired quality is achieved. Although each coil comes with a mill certificate (3.1 certificate), this information is not sufficient to set the straightening machine and fine-blanking tools correctly on the first attempt. The characteristics of individual coils vary too much. Within SPAICER, a digital twin of the coils will be created, containing more data, especially on microstructural properties, to make machine and tool adjustments more efficient in the future.

The digital coil

Feintool Systems Parts in Jena receives its steel coils from Reinhold Mendritzki Kaltwalzwerk GmbH & Co. KG. Currently, each coil is delivered with a mill certificate describing its characteristics, such as geometry and mechanical properties, and confirming compliance with customer specifications.

Typically, a coil’s characteristics include geometry, tensile tests, chemical composition, hardness, and microstructure. However, these represent only a small portion of a metal’s properties. Furthermore, measurements can only be taken at the start and end of a coil, as the tests are not non-destructive. As a result, downstream processes like straightening and fine-blanking require repeated adjustments for each coil change, and immediate desired production quality is not guaranteed.

Within SPAICER, additional magnetic measurements will be used to create a digital twin of the coil, the “Digital Coil”, providing much more informative data than the traditional mill certificate.
The aim is to develop a comprehensive AI-based process model using the digital coil data from Mendritzki and production data from Feintool, which automatically suggests optimal parameters for straightening and fine-blanking. This could allow production to start at 100% quality from the outset. deZem supports the development of the data infrastructure and AI modules.

Vision: New business models

Digital twins open up a whole new dimension in business model development. Digital data can be offered alongside physical products, adding value for customers. This is especially valuable in the processing industry, as it helps improve processes and production workflows while ensuring consistent product quality. Unnecessary setup times, test part production, costs, and material can be reduced. It is a win-win situation for suppliers and customers and lays the foundation for a transparent and trustworthy collaboration.