BMW Group AG / Site Steyr

Almost every second new BMW Group vehicle sold has a “heart”, i.e. a drive system from the Group’s largest engine plant in Steyr, Austria. There, 4,400 employees produce more than 1 million engines and around 12 million components in several product groups every year.

A few years ago, the plant launched its own flagship initiative of digitalization, which continues to deal with various aspects of digital transformation across all plants. This initiative is not intended to pilot new individual applications. Instead, it is designed to enable the entire plant to pursue and implement an overall approach to digital transformation.

In addition to the use of AI technologies, for example for empties management and quality inspection, the conceptual orientation of this winning project is therefore exemplary. It focuses on solving two tasks: what structural, organizational, and cultural change processes need to be set in motion in order to continuously integrate digitalization into the various departments at the plant? And how can the growth of digitalization solutions be further accelerated?

The plant team uses an exchange event, the so-called “digitalization platform”, to regularly collect new, department-specific digitalization topics. In addition, the topics and projects already initiated are discussed, adapted or further developed across departments within this framework. Three exemplary implementation aspects of the digital transformation at the plant are:

• Digital index: measurement and assessment of digital maturity in all plant departments, additionally this has been aggregated at plant level. This allows potential to be identified and continuous improvement to be achieved.

• Work organization: identification of ways to further develop the existing work organization. Among other things, the focus is on the effective use of new “Toolbox 4.0” methods such as AI, process mining and data analytics.

• Plant data access: definition of three focal points – value stream, quality and process efficiency – for cross-divisional data linking.

The interface for coordinating all digitalization activities in the plant is the “InnoLab”. Among other things, it provides a transparent overview of all internal plant solutions and technologies for the entire workforce in a digitalization map. However, the InnoLab’s tasks include not only the functional orchestration of around 200 Industry 4.0 technologies (as of 2023), but also the empowerment of all employees and site levels to use these tools and to use data in a value-added manner.

More than 100 different quality features have to be checked for each component type in the plant’s engine and component production. Using AI and machine learning, the award-winner identifies anomalies on the component and trains the AI with “in order” (IO) images. Production process influences such as cleaning residues or milling marks on component surfaces can have a negative impact on the optical properties without affecting the component quality. This in turn leads to significantly increased pseudo error rates.

With the help of a special anomaly detection algorithm developed in-house, the plant uses the “style transfer” principle to convert artificially generated images so that they look like real photos. This allows an infinite number of images of all possible error types to be generated, which in turn significantly reduces the pseudo error rates. Another advantage of this AI application is its scalability: the system works independently of the input data for different components such as crankcases, cylinder heads or electric motor housings.