Optimising Battery Aging Studies with ML: A Collaborative Approach

In a collaborative webinar, industry experts Dr. Gavin White, CEO and co-founder of  About:Energy, and Dr. Richard Alfeld, CEO and founder of Monolith, discussed the challenges of degradation in battery technologies and the innovative ways they are leveraging machine learning (ML) to optimise aging studies.  This blog will highlight some of the insights shared during the presentation.  




Key Challenges Discussed


  1. Resource Constraints: The industry faces a shortage of battery engineers, making it challenging to execute numerous projects effectively.
  1. Data Quality Issues: Poor-quality data from cell vendors or test labs complicates the understanding of battery behaviour during ageing.
  1. Cost and Timing: Conducting degradation studies is expensive, with OEM-level studies easily reaching millions of pounds. Moreover, obtaining results can take up to two years. 


Challenges in Battery Aging Studies



The session highlighted the significant challenges in battery aging studies, emphasising the high cost, time-consuming nature, and resource-intensive aspects of conducting such experiments.  

Dr. White outlined common issues such as the shortage of battery engineers, poor-quality data from cell vendors or test labs, and the inherent difficulty of designing effective test matrices that yield accurate results.  


Test Matrix Development


Dr. White of About:Energy emphasised the importance of developing a robust test matrix to gather valuable information efficiently.  

The collaboration with Imperial College London involves ongoing research to advise on degradation testing, enabling About:Energy to optimise experiments and gain insights from the data collected.  


Case Study: Battery Aging Data


Dr. White presented data from aging studies on various cells, including the Molly Cell P45B, LGM 50 Lt., and the lithium works LFP cell. The data showcased the impact of different aging conditions on capacity fade, highlighting the need for comprehensive testing matrices to understand battery behaviour accurately.  



Addressing Data Challenges


To tackle data challenges, About:Energy employs a rigorous approach to ensure data quality, including designing experiments with precise control over temperature and other variables.   

This commitment to data quality is crucial for making informed decisions based on battery degradation data 


Monolith's Contribution


Dr. Richard Alfeld from Monolith discussed how machine learning can enhance battery testing efficiency.   

Monolith specialises in empowering engineers with AI and ML tools to solve complex physics problems and accelerate product development.  



Machine Learning for Test Plan Optimisation


Dr. Alfeld introduced the concept of test plan optimisation using machine learning, focusing on techniques like forecasting, active learning, and early stopping to maximise learning during testing. 

He showcased how these tools, combined with high-quality data from About:Energy, can significantly reduce the time and resources needed for aging studies  



Early Stopping and Forecasting


The demonstration included a walkthrough of the Monolith platform, demonstrating how machine learning models can forecast battery behaviour and identify tests that may not yield valuable insights.  



This enables researchers to optimise their test matrices by stopping certain tests early, saving time and resources.  




Future Directions


The collaborative efforts between About:Energy and Monolith aim to revolutionise battery testing methodologies. By leveraging machine learning and high-quality data, the goal is to develop pre-trained models that provide accurate predictions, ultimately streamlining the product development process for battery technologies.  




The webinar highlighted the synergy between About:Energy and Monolith in addressing the challenges of battery aging studies. 

The integration of high-quality data, advanced machine learning techniques, and a commitment to data quality offer promising prospects for optimising battery testing and accelerating the development of innovative energy storage solutions. Stay tuned for more updates as this collaboration continues to evolve. 


About:Energy Overview  


About:Energy is a company that emerged from Imperial College London and the University of Birmingham. They specialise in constructing different types of battery models, such as electrical, electrochemical, thermal, and degradation models.  

These models are based on high-quality data obtained through rigorous testing of commercially available cells from leading manufacturers like Samsung, Panasonic, and Molly Cell. 

Share this post

Request a Demo

Subscribe to our monthly newsletter

Share this post

Ready to get started?

BAE Systems
Mercedes Benz