Smarter Charging Could Add Years to Your Battery’s Life

A charging algorithm developed at Chalmers University of Technology extended the simulated lifespan of a lithium-ion battery cell by nearly 23 percent compared with the standard protocol used in electric vehicles today, according to research published in IEEE Transactions on Transportation Electrification [1]. The controller, described in the paper “Lifelong Reinforcement Learning for Health-Aware Fast Charging of Lithium-ion Batteries,” adds less than three seconds to a typical charging session while potentially delivering two to three extra years of usable battery life [1]. The method applies only to Level 1 and Level 2 AC charging — the overnight home or workplace charging that covers most EV owners — and not to DC fast charging, which operates under different electrical constraints [1]. Rechargeable lithium-ion batteries, which power everything from smartphones to grid-scale storage, degrade with each charge-discharge cycle, and billions of dollars are invested globally in extending their longevity [4]. The standard industry protocol, constant-current constant-voltage charging, was designed around new cells and never updated to account for how aging raises internal resistance and narrows a cell’s safe operating window [1]. The Chalmers controller instead uses the battery’s real-time state of health as a live input, easing the charging current as capacity fades [1]. Meng Yuan, the paper’s lead author and now an assistant professor at Victoria University of Wellington, said that for fleet operators who charge vehicles multiple times daily, “a 23-percent life extension would meaningfully change their economics” [1]. The researchers used a reinforcement-learning technique that ran through thousands of simulated charging cycles to discover strategies that balance battery protection against charge time [1]. The resulting controller takes just two inputs — voltage and state of health — and outputs a current level, with all learning completed before deployment [1]. Yuan said the algorithm could reach existing vehicles as an over-the-air software update to their battery management systems, requiring no new hardware [1]. Tesla, the world’s leading battery-electric vehicle manufacturer with a 17.6 percent market share in 2024, and Volvo Cars, which owns a 50 percent stake in battery venture NOVO Energy, both routinely deliver vehicle improvements through over-the-air updates, a practice now common across the industry [6][7]. The work has so far been validated only on a single simulated cell at a constant 25 degrees Celsius, never on a physical battery or a multi-cell pack [1]. Yuan acknowledged that real-world temperature variation is an important variable and said follow-on work will adapt the controller to additional battery chemistries, which will require separate characterization for each [1]. Industry contacts have reached out since publication, but formal engagement with automakers or regulators has not yet begun [1]. The hardware testing needed to confirm or complicate the simulation’s findings will take months of physical cycling [1].