Master Thesis: Development of a predictive maintenance concept for cell component production of lithium-ion battery cells in the automotive sector
In cooperation with the institute for factory automation and production from the university of Erlangen-Nuremberg
Integrating predictive maintenance in the framework of lithium-ion battery manufacturing
MANAGEMENT SUMMARY
The increase in battery production including numerous gigafactories implies inevitable improvement in manufacturing efficiency and productivity. Due to this rapid development, integrated engineering systems become increasingly complex. This entails unanticipated faults in production processes that range from simple replacements to accidents, provoking fast costs in lost productivity as well as human and environmental resources. Damage induced or regular maintenance strategies according to the standard DIN 31051 do not exhibit capabilities in completely eradicating faults and thus no longer satisfy modern lithium-ion battery industry. Certainly, the reduction of downtime and maintenance expenses in the context of a zero-failure production is a critical aspect for a manufacturer in being sustainably competitive. Under the premise of a maximum achievable reliability in production equipment, a higher degree in automation of condition monitoring is desired. Therefore, this whitepaper investigates the potentials and trends of predictive maintenance strategies.
The development of predictive maintenance is presented in a systematic concept, its technical challenges, and the potentials under the premise of intelligent data acquisition and processing. The concept includes scenarios of sensor selection and data acquisition, data preprocessing in data mining and decision support for advisory maintenance implementation. This offers a holistic understanding and instruction in all domains for industry practitioners. From the authors perspective it is important to counteract the complexity in battery cell manufacturing. In a hybrid approach, critical components are focused to be simulated and diagnosed with deep learning tools to extract microstructural information and predict impending failures, respectively. However, also process data including e.g. intermediate product information, provide failure inducing aspects. The proposed methods and techniques should enable the feasibility of diagnostic and prognostics in the framework of maintenance and deal with crucial data mining aspects.
The potentials that can be gained by integrating predictive maintenance
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
