Elodie Collins 
SandboxAQ has completed a full-scale calendar-aging campaign to advance battery analytics using artificial intelligence and improve mission readiness at the U.S. Army.
The company said Monday that the campaign combined storage testing with deep-cycle operational data across multiple lithium-ion chemistries and form factors, creating an expansive dataset that represents a wide range of storage durations and temperatures. The dataset can enable the Army to better understand how temperature, state of health and operational profiles determine a battery’s capacity loss, charge retention and remaining useful life.
SandboxAQ Develops AI Model for Predicting Battery Condition
According to SandboxAQ, its latest generation of large quantitative models, or LQMs, can ingest a single reference performance test and predict a battery’s capacity fade over future storage scenario.
Based on preliminary internal evaluations, SandboxAQ’s LQMs predict how a battery’s performance will decline with a less than 1 percent error over 18-month horizons. The technology shortens testing timelines by as much as 90 percent and reduces waste by preventing premature battery disposal.
“The Army’s tactical edge depends on batteries that perform flawlessly after years on the shelf and months in the field,” commented Ty Sours, senior research scientist at SandboxAQ. “By marrying the Army’s extensive test matrix with our physics-informed LQMs, we can now deliver actionable health and logistics insights in days–not years–while slashing sustainment cost and risk.”
The project is a collaboration between SandboxAQ and the U.S. Army Combat Capabilities Development Command’s Command, Control, Communications, Computers, Cyber, Intelligence, Surveillance and Reconnaissance Center.
