One of the greatest challenges in energy storage is the long horizon required for the optimization and implementation of new materials and concepts at the commercial level. This is due to the massive multidimensional optimization parameter spaces as well as the long time required to test the performance lifetime of a given set of parameters. StorageX tackles this problem by bringing together faculty from materials science and data science to address the two major bottlenecks in battery development: navigating massive multidimensional optimization parameter spaces, and testing the performance lifetime of a given set of parameters.
Using advanced machine learning methods on a wide range of data streams, we look for diagnostic features that predict battery lifetime under a wide range of use cases and for a wide range of different battery chemistries. Aside from accelerating cell testing during development and optimization, these capabilities allow for:
- Adaptive charging algorithms to minimize certain types of degradation based on how and where a battery is used
- Grading batteries for second life applications
- Extending the warranty of commercial cells and packs
- Pricing and warrantying used electric vehicle battery packs for second-life applications
StorageX supports several projects aimed at developing this capability at the cell, module, and pack level involving faculty in Materials Science, Systems Engineering, and Business.
We develop advanced experimental design and machine learning algorithms as well as physical infrastructure to automate and accelerate the development and testing processes. This includes efficiently utilizing hundreds of parallel battery testing channels and leveraging performance prediction models to rapidly navigate massive complex parameter spaces.