The most expensive part of your EV may depend on the driver's right foot

Researchers at Shanghai Dianji University analysed a year of real-world data from 15 electric vehicles operating in Guangzhou in 2022 and concluded that driving behaviour itself is a major factor in how fast a battery ages ^[1]. Feeding each driving style into a battery-aging model, they projected that the most aggressive style would lose 53.22% of original capacity after 1,000 charge-discharge cycles, against 21.15% for the smoothest ^[2] - roughly 2.5 times the degradation, or about 32 percentage points of usable capacity ^[3][4].

The definition of "aggressive" is the part worth reading closely. It was not high speed ^[5]. It was repeated demands for high electrical current: frequent high-torque acceleration at relatively low speeds, the traffic-light launch repeated all day ^[5]. A car cruising fast on a highway is not necessarily doing the same thing to its pack as one that launches, brakes, and launches again ^[6]. That distinction matters for anyone building driver-scoring products, because speed is the easiest signal to collect and appears to be the wrong one.

The current numbers are the mechanism. The aggressive group averaged 66.02 amps root-mean-square; the eco group 20.56 ^[7]. That is a 3.2x gap in current ^[8], but the study reports the aggressive category produced an immediate aging stress rate 18.4 times higher than eco-driving ^[9]. Stress, in other words, does not scale linearly with the current drawn, which is why "drive a bit more gently" is not a proportional intervention.

Two things temper the headline. First, the arithmetic is not internally tidy: when the researchers modelled changing only driving behaviour from aggressive to eco while holding other conditions constant, short-term aging stress fell by 91.3% ^[10], which implies roughly an 11.5x reduction ^[11], not the 18.4x gap between the raw categories. That is the difference between comparing two populations and isolating one variable, and it is the more defensible of the two figures.

Second, nobody watched a car for 1,000 cycles. The team did not directly observe the same vehicles over 1,000 charge cycles and measure the resulting capacity loss; they estimated stress from real-world data and applied an aging model ^[12]. The 53.22% and 21.15% figures are modelled projections, not measurements ^[13], and real degradation also depends on charging patterns, climate, chemistry, vehicle weight, terrain, traffic, and thermal management ^[14]. Fifteen vehicles, one city, one year ^[1] is a direction of travel, not a coefficient to underwrite anything with.

The commercial reason to care anyway: the pack is one of the most expensive components in an EV, and lost capacity shows up directly as lost range, plus effects on performance and resale value ^[15]. Residual value models and battery-health warranties are currently priced on things the operator can see - mileage, age, charging behaviour. This study argues that a large, cheap lever sits somewhere else entirely, in torque mapping, pedal calibration, and driver coaching, and that current-based telematics would reveal it.

What to watch: whether anyone replicates this on a fleet large enough to separate driver from route, whether leasing and rental firms start scoring current draw rather than speed and harsh-braking events, and whether OEMs quietly soften low-speed torque delivery in software while continuing to advertise the launch times that produce exactly the load pattern the study flags ^[5].