Making companies more resilient

A new project supports companies on using Artificial Intelligence (AI) to survive economically critical situations stably.

A new project supports companies on using Artificial Intelligence (AI) to survive economically critical situations stably.

Crises show that when only one link in the supply chain fails in our digitalized high-tech world, this means high risks and costs for companies. The SPAICER project launched in April 2020 by the German Research Center for Artificial Intelligence (DFKI) and Saar University together with partners from science and industry seeks to ensure a plan B and the corresponding “resilience management.” As a manufacturer of specialty glass, SCHOTT is a partner in this beacon project that is being supported by the German Federal Ministry of Economics and participating companies to the tune of a total of more than EUR 10 million. What is the project about? SPAICER stands for “Scalable adaptive production systems through AI-based resilience optimization.” The idea is to develop a data system that uses AI to enable companies to detect potential production disturbances early on, react to them in the best possible way and thus make them resistant respectively resilient to crises. The project is thus also intended to help strengthen the German economy for international competition and to make it competitive with the help of AI technology.

“For us, participation in the project is a great opportunity to learn more through cooperation with various leading research institutions in the field of AI. At SCHOTT, we have various projects in which we use AI-based methods for improving our manufacturing processes,” explains Dr. Markus van Ackeren, Head of Data Science at SCHOTT Research. But why is AI so essential, particularly in glass production? Extremely pure raw materials, such as rare earths (including lanthanum), are needed for optical applications, for example. And a glass melting tank in which raw materials are melted at temperatures of over 1000 °C cannot be temporarily shut down. Dr. van Ackeren explains: “Security of supply of raw materials, therefore, plays a decisive role.” However, due to weather-related restrictions, e.g., when transporting by ship, the availability of raw materials can deteriorate drastically. Interruptions in the supply chain then lead to lost sales and put pressure on customer relations. SPAICER optimizes production planning based on forecasted water levels, temperature developments and the processing of other signals such as holiday periods and trends in the logistics industry by stabilizing the supply chain, for example, through early, alternative logistics planning via roads, rail and waterways as well as requests from alternative suppliers of the required raw materials.

Furthermore, the SPAICER system is intended to make the probable effects of global events on production transparent and to provide recommendations for action in the event of political crises for optimized planning. To this end, trend data for raw material prices and analyses of political developments, for example, are continuously fed into the system to predict disruptions such as the loss of suppliers. The data is then used to develop solution proposals with the help of AI-based algorithms. SPAICER is also intended to make companies fit for unforeseen internal disturbances, such as plant failures, for example, by using AI and various machine learning methods to classify disturbances and recommend appropriate measures. For SCHOTT and the other project partners, the SPAICER project, therefore, represents an essential building block in solving the question of how to make production resistant to current and future crises.

Dr. Markus van Ackeren is Head of Data Science in the central research and development department at SCHOTT. Together with his team, the expert for data analysis and machine learning develops mathematical methods to predict how components will deteriorate in hot processes as part of SPAICER. This form of predictive maintenance is intended to prevent failures and make measures more predictable.


Dr. Markus van Ackeren
Research and Development