There’s principle after which there’s actuality. The connection between them can usually be tenuous at finest. On this planet of electrical automobiles, every battery pack has a battery administration system that makes use of algorithms to observe its general well being and working situation. “The algorithm tells you issues like in case your battery is doing okay, or how far you may drive earlier than it’s essential recharge. The issue is that BMS algorithms are designed in preferrred laboratory circumstances that don’t mirror what a battery pack sees in the actual world,” says Simona Onori, an assistant professor of vitality science and engineering within the Stanford Doerr Faculty of Sustainability.
To reveal the hole between managed laboratory testing and precise highway expertise, Onori and colleagues at Stanford collaborated with researchers on the Volkswagen Innovation and Engineering Heart positioned close to the college campus. “Algorithms primarily based on unrealistic driving information are prone to be inaccurate within the discipline,” mentioned Onori, lead creator of the research. “Our purpose is to extend the longevity of the battery pack by designing algorithms skilled from real-world information.”
The outcomes of the analysis had been printed on August 18, 2023 within the scientific journal Joule (paywall). Here’s a abstract of that report:
“Deploying battery state of well being estimation and forecasting algorithms are essential for making certain the dependable efficiency of battery electrical automobiles. SoH algorithms are designed and skilled from information collected within the laboratory upon biking cells below predefined hundreds and temperatures.
“Area battery pack information collected over 1 yr of auto operation are used to outline and extract efficiency/well being indicators and correlate them to actual driving traits (charging habits, acceleration, and braking) and season-dependent ambient temperature. Efficiency indicators throughout driving and charging occasions are outlined upon establishing an information pipeline to extract key battery administration system indicators.
“This work exhibits the misalignment current between laboratory testing and precise battery utilization, and the chance that exists in enhancing battery experimental testing to deconvolute time and temperature to enhance SoH estimation methods.”
What A Battery Administration System Does
Picture credit score: Stanford College through Joule
A battery administration system at present routinely information information throughout braking, acceleration, deceleration, and charging. The key to lengthy life for rechargeable batteries might lie in understanding the variations between particular person cells in actual world operation. New modeling of how lithium-ion cells in a pack degrade exhibits a strategy to tailor charging to every cell’s capability so EV batteries can deal with extra cost cycles and an extended service life.
“Actual-world driving is driver specific,” says co-author Gabriele Pozzato, a Stanford analysis engineer. “You could be an aggressive driver, or somebody who solely partially costs their automobile. Totally different types of driving and charging will end in totally different trajectories of battery degradation. Nonetheless, that sort of discipline information shouldn’t be included in standard battery algorithms.”
For the research, Volkswagen offered the Stanford workforce about 3,750 hours of BMS driving information collected from an all-electric Audi e-tron SUV pushed within the San Francisco Bay Space for one yr, from November 2019 to October 2020. That information allowed the Stanford workforce to calculate {the electrical} resistance within the battery pack throughout that interval. The information allowed the workforce to evaluate two key battery metrics — vitality and energy.
“Vitality offers you the vary, or what number of miles you may drive with a totally charged battery,” Pozzato says. “Energy is the flexibility to extract vitality rapidly. While you speed up, you wish to get entry to vitality and discharge the battery very quick. The much less electrical resistance you’ve got within the battery, the extra energy you’ve got.”
To calculate resistance, the researchers measured abrupt modifications in present and voltage within the battery pack utilizing information from 529 acceleration occasions and 392 braking occasions in the course of the yr. In addition they calculated impedance — a measure of resistance throughout battery charging — by analyzing 53 charging classes. “Impedance and resistance are usually thought of metrics of battery well being,” Onori mentioned. “The extra you drive, the extra resistance will increase. This normally interprets into much less out there energy from the battery pack, however that’s not what we noticed.”
A extra complicated sample emerged when the researchers added seasonal climate information to the combination. They found {that electrical} resistance decreased in cooler months and steadily elevated in spring and summer time, a sign that battery well being improves as temperatures rise.
“Larger temperatures increase battery capability, so you’ve got this sense that the automobile has extra vitality and which you can drive extra miles,” Onori mentioned. “However for those who maintain utilizing the battery at excessive temperatures, it’s going to degrade sooner. These are very difficult elements that have an effect on efficiency. Subsequent yr we’ll develop our dataset to a fleet of automobiles to find out precisely how temperature and getting old have an effect on one another.”
Idea Vs. Actuality
Automakers depend on standard battery administration algorithms designed in preferrred laboratory circumstances. Utilizing machine studying, these algorithms usually monitor efficiency information from a single 4-volt battery cell that repeatedly costs and discharges at a relentless temperature till it dies. However the Audi discipline information was collected from a 396-volt battery pack powered by 384 cells.
“New algorithms ought to concentrate on the complete battery pack and never particular person cells,” Onori says. “We wish to design algorithms that educate drivers on how you can enhance the lifetime of the battery pack, which is the most costly part of the car. For instance, you can alert drivers if they’re fast-charging an excessive amount of or accelerating too aggressively. A lot may be realized from discipline information to make BMS algorithms extra strong.”
The research was assisted by members of the Volkswagen Innovation and Engineering Heart in Belmont, California, only a hop, skip, and a soar away from the Stanford campus. Computational sources had been offered by the Stanford Analysis Computing Heart. The authors have filed three patent functions associated to this work.
The Takeaway
This analysis is an instance of how new info can result in new insights into how you can maximize the efficiency of battery packs for electrical automobiles. The EV period at present is about the place the transition to inner combustion engines was a century in the past. Whereas the idea is that the engine in a 1923 Ford operated on the identical fundamental ideas because the engine in a 2023 Ford F-150 — usually lowered to the extremely simplistic phrase “Suck, Push, Bang, Blow” — the truth is the fashionable engine is gentle years forward of its predecessor when it comes to the extracting the utmost quantity of vitality from a drop of gasoline.
EV expertise is following an identical path of fixed enchancment. Research similar to this one at Stanford will assist future electrical automobiles maximize the efficiency and longevity of their battery packs. A greater battery administration system could also be essential to creating it potential for EVs to export their saved vitality to energy exterior hundreds with out shortening their helpful life.
Theories are nice beginning factors, however there isn’t a substitute for actual world expertise.
I do not like paywalls. You do not like paywalls. Who likes paywalls? Right here at CleanTechnica, we carried out a restricted paywall for some time, nevertheless it all the time felt improper — and it was all the time robust to resolve what we should always put behind there. In principle, your most unique and finest content material goes behind a paywall. However then fewer individuals learn it! We simply do not like paywalls, and so we have determined to ditch ours.
Sadly, the media enterprise continues to be a troublesome, cut-throat enterprise with tiny margins. It is a endless Olympic problem to remain above water and even maybe — gasp — develop. So …
