Batteries News

SandboxAQ’s Large Quantitative Models (LQMs) Help Decrease Battery Life Prediction Time by 95%

Battery Life Prediction Time

SandboxAQ’s Large Quantitative Models (LQMs) Help Decrease Battery Life Prediction Time by 95%

LQMs trained on NOVONIX Ultra-High Precision Coulometry data predict battery life with 35x greater accuracy and 50x less data than traditional AI models

PALO ALTO, Calif., Oct. 8, 2024 /PRNewswire/ — SandboxAQ today announced that its Large Quantitative Models (LQMs) reduced the time needed to predict lithium-ion battery end-of-life (EOL) by 95%, with an unprecedented 35x greater accuracy and 50x less data than traditional approaches – reducing the time for cell testing from months or years to just days. This breakthrough was achieved by training SandboxAQ’s LQMs on more than 4 million hours of Ultra-High Precision Coulometry (UHPC) cycle data collected over five years from internal development projects by leading battery materials and technology company NOVONIX Limited (NOVONIX). The two companies will discuss the results via presentations at The Battery Show this week and a journal paper.

Traditionally, new battery designs require 5-10 years of costly experimentation and testing to discover and optimize energy density, longevity, safety, and rate capability with 90% of designs never reaching commercial development. Leveraging physics-based NOVONIX UHPC data, SandboxAQ’s LQMs and Quantitative AI simulations can capture subtle changes in charge-loss and efficiency variations across early battery cycles to gain nuanced insights into the electrochemical processes that drive long-term cell degradation. This approach shows significant promise for accurate cycle life prediction, generalizability across previously untested cell chemistries, and cycling conditions at different charge rates, cut-off voltages, and temperatures.

Initial applications of SandboxAQ LQMs indicate that they can potentially shave two to four years off the cell’s development and commercialization timeline and save cell manufacturers millions of dollars in R&D costs. These savings translate into a faster innovation cycle, enabling the advancement of battery technology across multiple industries and accelerating the adoption of new solutions to meet the growing demand for high-performance energy storage.

Dr. Ang Xiao, Technical Lead, AI for Materials Science at SandboxAQ, said :

SandboxAQ’s Large Quantitative Models continue to demonstrate significant value across industries by significantly lowering the time, cost and risk of experimental lab R&D, improving processes, and delivering new insights that lead to innovative new products,

“The results from our collaboration with NOVONIX will help speed more effective and eco-friendly batteries and energy storage solutions to market for a broad range of commercial, industrial and public sector applications.”

SandboxAQ’s LQMs, paired with NOVONIX UHPC cyclers, accurately predicted cell EOL within a mean absolute error (MAE) of 11 cycles using just 40 cycles of UHPC data. For a dataset of NMC-based cells spanning three manufacturers, the UHPC-based model showcased a 35x increase in accuracy compared to traditional capacity-based models using features from standard cyclers. Moreover, with only two cycles of UHPC data, SandboxAQ predicted EOL with a MAE of 50 cycles, representing a 50x faster time-to-prediction compared to traditional models requiring 100 cycles of data.

Dr. Stephen Glazier, Director of Technology at NOVONIX, Battery Technology Solutions Division, said:

Predicting the cycle life of lithium-ion and other battery cells is inherently difficult due to the limited availability of high-quality, large-scale datasets and the lack of generalizable feature sets that can accurately model cell degradation,

“SandboxAQ’s novel approach of combining LQMs with our comprehensive UHPC data to accurately predict long-term performance could have significant implications on future cell development by helping manufacturers make better informed decisions ranging from optimizing designs, materials and chemistries, to warranty and time-to-spec predictions, to QA/QC metrics in manufacturing, and more.”

In July, the U.S. Army Futures Command’s Command, Control, Communications, Computers, Cyber, Intelligence, Surveillance and Reconnaissance Center (C5ISR Center) selected SandboxAQ to help support its power and energy modernization initiatives. Since then, the organizations have been collaborating to develop advanced battery chemistries and designs for diverse applications such as electric vehicles (EVs), Unmanned Aerial Vehicles (UAVs), and portable power solutions. This initiative seeks to improve battery performance, range, safety, fast-charging capabilities, and integration with existing infrastructure while mitigating costs and environmental impact. 

The Battery Show NA (Detroit, MI; Oct. 8-10) attendees can meet with leaders from SandboxAQ (Booth #4245) and NOVONIX (Booth #3208) to learn how their technologies are accelerating the development and testing of innovative new battery chemistries, materials and designs. The results will be highlighted in presentations by NOVONIX (“Applications of NOVONIX Ultra-High Precision Coulometry Across the Battery Supply Chain;” Oct. 9, 11:00am ET), and SandboxAQ (“Discover Tomorrow’s Battery Materials Today with Large Quantitative Models;” Oct. 10, 2:30pm ET) at the show.

For more information about SandboxAQ’s LQMs, please visit https://www.sandboxaq.com/solutions/large-quantitative-models. For more information about NOVONIX UHPC technology, please visit https://www.novonixgroup.com/battery-technology-solutions/uhpc.

READ the latest Batteries News shaping the battery market

SandboxAQ’s Large Quantitative Models (LQMs) Help Decrease Battery Life Prediction Time by 95%, source

batteries news

Get our LinkedIn updates!

Join our weekly newsletter!

Follow us

Don't be shy, get in touch. We love meeting interesting people and making new friends.