The race for longer-lasting batteries is accelerating advancements in consumer electronics, electric vehicles, and renewable energy storage. However, the slow and costly process of cycle life testing remains a major obstacle, delaying breakthroughs.
SandboxAQ is transforming battery research by integrating artificial intelligence (AI) with multiscale physics simulations. Through Large Quantitative Models (LQMs) and synthetic data from atomistic simulations, we are uncovering deeper insights into battery performance and degradation—at unprecedented speed.
In collaboration with NOVONIX, we have developed generalizable cycle life prediction models that address the challenges of predicting battery longevity across different chemistries and manufacturers. These models offer a robust, scalable solution to streamline battery development and deployment, making the process more efficient and reliable.
Cycle life testing involves repeatedly charging and discharging batteries until they degrade, a process that can take weeks, months, or even years depending on the chemistry and usage conditions. This becomes increasingly complex when multiple batteries need to be tested under various conditions to ensure real-world performance. Traditional testing methods are not scalable for the rapid development cycles demanded by today's market.
Machine learning models offer a promising solution by predicting end-of-life (EOL) without the need for exhaustive testing, potentially reducing testing times from years to days. However, creating models that accurately predict EOL across different battery chemistries and manufacturers is challenging. Battery performance varies widely due to factors like electrode materials, cell design, and manufacturing processes. Limited data availability further complicates model development, as collecting extensive datasets is both time-consuming and expensive.
To tackle these challenges, SandboxAQ partnered with NOVONIX, a leader in battery analytics technology. Our joint effort focused on developing generalizable cycle life prediction models capable of delivering accurate predictions even with limited data.
By leveraging NOVONIX’s Ultra High Precision Coulometry (UHPC) systems, novel Direct Current Internal Resistance (DCIR) measurements, and high-resolution differential capacitance (dQdV) analysis, we have developed cutting-edge feature sets and model architectures that uncover the subtle, often elusive degradation mechanisms leading to battery failure. This breakthrough enables our proprietary model to accurately predict cycle life across previously unseen manufacturers and chemistries—a transformative leap in battery research that redefines the speed and precision of innovation. Our novel AI solution provides valuable insight into the complex underlying degradation pathways, improving our understanding of the factors influencing degradation and guiding manufacturers toward optimized performance.
Generalizable models are a catalyst for battery innovation, enabling faster, more efficient development across the industry. By allowing original equipment manufacturers (OEMs) to leverage existing data, these models reduce the need for costly re-certification and extensive cycling tests whenever battery designs or materials are modified. This adaptability fosters an agile development process, accelerating the path from concept to commercialization.
Our work proves that with strategic feature engineering and advanced model development, it's possible to overcome data scarcity and build robust, high-performing models. Identifying the key indicators of battery degradation not only enhances prediction accuracy but also unlocks critical insights for material optimization, performance improvements, and greater reliability.
At SandboxAQ, we are committed to pushing the boundaries of battery research—and our collaboration with NOVONIX is just the beginning. By fusing AI, multiscale physics simulations, and quantum mechanics, we are developing Large Quantitative Models (LQMs) that connect atomic-scale interactions to real-world performance metrics. This comprehensive approach allows us to build more accurate, adaptable, and generalizable models, capable of predicting critical performance indicators across a diverse range of battery chemistries and operating conditions—accelerating innovation and driving the next generation of energy storage solutions.
Stay tuned for more insights on how quantitative AI and Large Quantitative Models (LQMs) are transforming the future of battery technology and visit our website for more information about AQChemSim.
AUTHORS:
Dr. Tyler Sours is a senior researcher in the AI Simulation Group at SandboxAQ. He has a Ph.D. in chemical engineering from the University of California, Davis, where he studied the application of multiscale atomistic models for materials discovery. At SandboxAQ, Tyler is focused on developing tools to integrate AI and simulation with experimental data to accelerate materials and process optimization.
Dr. Omar Allam is a scientist in the AI Sim Group at SandboxAQ. He holds a Ph.D. in Mechanical Engineering from the Georgia Institute of Technology, where he focused on integrating multiscale atomistic modeling and machine learning for enhancing energy storage and conversion materials. At SandboxAQ, Omar develops AI-enabled computational workflows that bridge quantum mechanics and machine learning to address complex challenges in materials science.
