AI-Driven Quality Control Aims To Cut Waste And Improve Safety In Battery Cell Production

Battery cell production is facing major challenges as both new and existing factories generate a high amount of waste. While electrical testing methods used today can measure basic performance parameters, they often fail to detect safety-related defects hidden inside battery cells. These undetected issues can affect battery life, safety, and sustainability.

To address this gap, an independent junior research group called NADINE has been launched to improve non-destructive testing methods for battery cells. The group is led by Dr. Moritz Kroll and has received around 2.3 million euros in funding from the German Federal Ministry of Research, Technology, and Space. The aim is to develop advanced quality assurance methods that can support competitive battery cell production in Europe while enabling safer and longer-lasting batteries.

More accurate quality control can deliver strong environmental and economic benefits. According to estimates by Fraunhofer ISE, lowering the reject rate by just one percentage point at a 10 GWh battery production plant could reduce carbon dioxide emissions by about 5,000 tons every year. It would also help save critical raw materials such as lithium and cobalt. The key factor is not only detecting defects, but also identifying their exact location and type inside the battery cell. This information can help manufacturers understand which production steps are responsible for defects and adjust their processes quickly and effectively.

The research project brings together battery research, data science, and artificial intelligence. It involves collaboration with researchers from TU Dresden and the University of Auckland in New Zealand. The team will focus on promising techniques such as ultrasound tomography and studying how battery cells expand during operation. By digitally connecting data from different stages of battery development and production, researchers can analyze links between manufacturing parameters, battery operation, health condition, and safety behavior.

Machine learning tools will be used to uncover new patterns and relationships that are difficult to detect using traditional methods. In line with FAIR data principles, which promote data that is easy to find, access, share, and reuse, the project will follow an open science approach and make its data sets available for wider research use.

Another important area of work is the evaluation of batteries for second-life applications. Batteries used in electric vehicles can still be useful for stationary energy storage if their condition is properly assessed. Developing reliable and fast testing methods can support battery reuse, strengthen the circular economy, and reduce the overall carbon footprint of energy storage systems.

The five-member junior research group will be based at Fraunhofer ISE’s Center for Electrical Energy Storage in Freiburg and will be integrated with the existing team there. The institute will provide access to its research infrastructure, including facilities for safe battery cell production and testing. An industrial advisory board will guide the project to ensure that the results remain relevant for industry and can be applied in German manufacturing.

The initiative also strengthens cooperation between the University of Freiburg and Fraunhofer ISE. Experts involved say that combining battery research with data science is an important step toward developing sustainable energy storage systems and practical solutions for industry.