A multi-stage lithium-ion battery aging dataset using various experimental design methodologies

Area of Science:

• Materials Science
• Electrochemistry
• Data Science

Background:

• Lithium-ion batteries are crucial for energy storage, but their aging mechanisms under combined stress are complex.
• Existing studies often rely on limited experimental designs, potentially overlooking critical degradation factors.

Purpose of the Study:

• To comprehensively investigate combined calendar and cycle aging in lithium-ion battery cells.
• To validate the efficacy of optimal experimental design (OED) methodologies in battery research.
• To generate a rich dataset for diverse battery performance and degradation studies.

Main Methods:

• Utilized 279 Samsung INR21700-50E lithium-ion cells under 71 distinct aging conditions.
• Employed non-model-based (full-factorial, Latin hypercube) and model-based (parameter individual optimal experimental design - pi-OED) approaches.
• Conducted a two-stage aging process to systematically explore degradation behaviors.

Main Results:

• Generated a comprehensive dataset on lithium-ion battery aging under combined stresses.
• Demonstrated the utility of OED in refining the understanding of battery degradation.
• Identified potential for uncovering previously hidden dependencies in battery aging.

Conclusions:

• The generated dataset is valuable for machine learning model training and physics-based model calibration.
• Optimal experimental design enhances the efficiency and depth of battery aging investigations.
• This research provides a foundation for more accurate battery performance prediction and improved battery management systems.