⚡ electrical engineering

Battery health reporting fails independent validation across manufacturers

This study reveals that on-board battery management systems across five major electric vehicle manufacturers fail to reliably report state-of-health compared to independent measurements, whereas a manufacturer-independent electrochemical marker achieves high accuracy in tracking battery degradation without relying on proprietary BMS data.

Original authors: Jeongju Park, Kyungkak Kim, Seungho Geum, Junhyung Lee, Hyeongyu Son, Sekyung Han

Published 2026-03-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jeongju Park, Kyungkak Kim, Seungho Geum, Junhyung Lee, Hyeongyu Son, Sekyung Han

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you bought a used car, and the seller handed you a digital dashboard that says, "Battery Health: 95%." You feel great! You think, "This car is practically new; I'm getting a great deal."

Now, imagine that same dashboard is actually a magic trick.

That's the core finding of this new study. Researchers looked at over 1,100 electric vehicles (EVs) from five different major brands (like Hyundai, Kia, Audi, and VW) and discovered that the "Battery Health" number your car shows you is often completely unreliable.

Here is the breakdown of what they found, using simple analogies:

1. The "Broken Thermometer" Problem

Think of your car's battery health report like a thermometer. If you have a fever, a good thermometer tells you exactly how hot you are.

The Reality: The researchers found that for many car brands, the "thermometer" is broken.
The Evidence: They found cars of the exact same model where the actual battery capacity differed by 12% to 25%. That's a huge difference! One car might have the range of a compact sedan, while its identical twin has the range of a small truck.
The Lie: Despite this massive difference, the car's computer (the BMS) often reported both cars as having "100% Health." It's like a thermometer reading "98.6°F" for both a healthy person and someone with a high fever.

2. The "Secret Menu" of Battery Health

The study found that the reliability of this health report depends entirely on who made the car, and sometimes even which specific model they made.

The "Blind" Cars: Some brands (like certain Audi and VW models) don't even show you the battery health number at all. It's like buying a house and the seller refusing to tell you the square footage or the age of the roof.
The "Random" Cars: For other brands, the number they show you is basically a guess. The researchers found almost zero connection between what the car says its health is and what the battery can actually do.
The "Inconsistent" Car Maker: Even within the same company (Hyundai/Kia), one model might give a somewhat accurate report, while a slightly different model from the same factory gives a completely random one. It's like a bakery where one baker makes perfect bread, but the one next door makes bread that looks perfect but is actually made of sawdust.

3. The "Lemon Market" Trap

This creates a dangerous situation for buyers, similar to the famous "Market for Lemons" theory in economics.

The Scenario: Imagine you are buying a used EV. The seller says, "My battery is 95% healthy!" You buy it.
The Twist: Because the seller's car is lying (or the car is lying), you might actually be buying a car with a battery that is only 70% healthy.
The Consequence: You can't tell the difference. The "good" sellers (who actually have healthy batteries) can't prove it, so they get paid the same as the "bad" sellers. Eventually, only the "bad" cars stay on the market because no one trusts the numbers. This hurts everyone: buyers get ripped off, and honest sellers can't sell their good cars.

4. The Warranty Loophole

This is the most frustrating part for owners.

The Promise: Most EV warranties say, "If your battery drops below 70% health, we will fix or replace it for free."
The Loophole: If your battery actually drops to 60% (which is bad!), but the car's computer still says "99%," you get no warranty help. The car is the judge, the jury, and the executioner. It decides when it is broken, and it rarely decides that it is broken.

5. The "Real" Health Check (The Detective's Tool)

So, how do we fix this? The researchers didn't just complain; they built a better tool.

The Old Way: Asking the car, "How are you?" (The car lies).
The New Way: The researchers developed a method that acts like a forensic detective. Instead of asking the car, they look at how the car charges.
The Analogy: Imagine you want to know if a runner is tired. You don't ask them; you watch how they run up a hill. If they slow down significantly at a specific point, you know they are tired.
The Solution: The researchers found that by looking at the voltage during slow charging, they could spot a specific "peak" that shifts as the battery gets old. This method works across all brands, doesn't need the car's secret software, and is 74% to 89% accurate.

Why Should You Care?

This study is a wake-up call for three big groups:

Regulators: The EU and California are planning new laws requiring battery health reports. This study says, "Hey, the current reports are garbage. You need to demand a standardized, independent test, not just a number from the manufacturer."
Used Car Buyers: Don't trust the dashboard number. If you buy a used EV, you need an independent check-up, just like you would with a used house or a used human.
Manufacturers: They need to stop hiding behind "proprietary algorithms" and start being transparent. If they can't prove their battery health numbers are real, they shouldn't be allowed to use them for warranties.

In short: The "Battery Health" percentage on your dashboard is currently a marketing gimmick, not a scientific fact. Until we have a universal, independent way to measure battery life, buying a used EV is a bit like buying a lottery ticket where the seller holds the winning numbers.

1. Problem Statement

The primary metric used by electric vehicle (EV) owners, regulators, and the used-car market to assess battery condition is the State of Health (SOH) reported by the on-board Battery Management System (BMS). However, this metric suffers from critical flaws:

Proprietary Opacity: BMS SOH algorithms are manufacturer-specific, undocumented, and lack standardization. A reported "95% SOH" may represent vastly different physical conditions across different brands.
Information Asymmetry: This creates a "market for lemons" scenario where the manufacturer (who holds warranty liability) is also the sole arbiter of when that liability is triggered.
Regulatory Gap: Upcoming regulations (EU Battery Regulation 2027, California ACC II) mandate SOH transparency, implicitly assuming reported SOH is reliable. No prior study has systematically validated BMS SOH against independent, ground-truth measurements across a large, real-world fleet.

2. Methodology

The authors conducted an epidemiological study of 1,114 EVs from five manufacturers (Hyundai, Kia, Genesis, Audi, Volkswagen) spanning 375 days of continuous telematics data.

A. Independent Capacity Measurement (The "Ground Truth")
Instead of relying on BMS data, the team developed a BMS-independent capacity measurement protocol based on Differential Capacity Analysis (dQ):

Data Source: Identified Constant-Current (CC) slow charging sessions from telematics data.
Strict Controls: Sessions were filtered for current stability (CV < 25%), temperature (10–35°C), and voltage window alignment (0.1 V/cell steps).
Metric: Accumulated charge ($dQ$) over standardized voltage windows serves as a proxy for actual capacity.
Validation:
- Multi-window consistency: Verified that $dQ$ ratios remained constant across different voltage windows.
- Ground truth: Compared narrow-window $dQ$ against wide-window ( $\ge$ 0.25 V/cell) measurements.
- Lab validation: Tested on an SK Innovation 55.6 Ah NCM pouch cell (identical to fleet cells) over 198 cycles. The partial-window $dQ$ tracked full-reference capacity with $\rho > 0.80$ .

B. Comparative Analysis
The study compared the independent $dQ$-derived capacity against the manufacturer-reported BMS SOH for each vehicle.

C. Usage Pattern Analysis
The authors utilized a Weighted Degradation Factor (WDF) approach, combining Principal Component Analysis (PCA) of usage patterns (SOC distribution, temperature, C-rate) with Ridge regression to determine what drives degradation variance beyond simple mileage.

D. Universal Markers
They investigated $V_{peak}$ (the voltage position of the dominant $dQ/dV$ peak) as a manufacturer-independent electrochemical marker for degradation.

3. Key Contributions

First Cross-Manufacturer Audit: The first independent validation of BMS SOH reliability against a single, standardized measurement protocol across a large real-world fleet.
Quantification of Information Asymmetry: Demonstrated that BMS SOH is fundamentally unreliable, failing to track real capacity differences in most models.
Field-Validated Universal Indicators: Proved that $V_{peak}$ (an electrochemical feature) and usage patterns can serve as manufacturer-independent health indicators without requiring BMS data.

4. Key Results

A. BMS SOH is Fundamentally Unreliable

Real Capacity Variance: Significant capacity heterogeneity exists within the same model (e.g., a 24.7% gap in commercial vehicles, 16.4% in E-GMP 192S).
BMS Failure to Track:
- Commercial (Porter/Bongo): 24.7% capacity gap resulted in a negligible 0.4% SOH gap ( $\rho = 0.24$ , non-significant). Most reported 100% SOH regardless of health.
- E-GMP (192S): 16.4% capacity gap resulted in a 1.0% SOH gap ( $\rho = 0.10$ , non-significant).
- Niro/Kona: 20.7% capacity gap resulted in a 1.6% SOH gap ( $\rho = 0.17$ , non-significant).
- MEB (Audi/VW): 384 vehicles (38% of the fleet) did not expose SOH at all via OBD-II, despite likely computing it internally.
- E-GMP 180S (Best Case): Showed moderate correlation ( $\rho = 0.62$ ) only under strict filtering ( $n=19$ ). In the unfiltered sample ( $n=63$ ), the correlation reversed to negative ( $\rho = -0.18$ ), proving the correlation is unstable and not robust.
Warranty Implications: Among the worst 10% of vehicles by actual capacity, 93% were not identified as degraded by the BMS. Vehicles reported as "healthy" (SOH $\ge$ 95%) had actual capacities ranging from 71% to 142%.

B. Usage Patterns Drive Degradation

Mileage is a Poor Proxy: Mileage alone explained negligible variance in health for most platforms ( $R^2 \le 0.28$ ).
SOC Dominance: Time spent at high State of Charge (SOC > 80%) was the primary driver of degradation. High-SOC users showed 10–13% worse energy efficiency than low-SOC users at the same mileage.
Temperature/C-Rate: Surprisingly, ambient temperature and fast-charging exposure showed weak or inconsistent associations with health in real-world conditions, likely due to effective thermal management systems.

C. Universal, BMS-Independent Markers

$V_{peak}$ Threshold: The voltage position of the $dQ/dV$ peak shifts from ~3.9–4.1 V (healthy) to ~3.6–3.7 V (degraded).
Classification Accuracy: A universal threshold of 3.67 V achieved 74–89% accuracy in classifying vehicles as degraded or healthy across all NMC platforms.
Augmented Models: Combining usage patterns with $V_{peak}$ significantly improved health prediction for platforms where usage alone was insufficient (e.g., MEB platform $R^2$ increased from 0.40 to 0.62).

5. Significance and Implications

Consumer Protection: Current warranty systems based on BMS SOH are structurally flawed, systematically failing to identify degraded batteries and denying claims to owners of "hidden" degraded vehicles.
Used EV Market: The lack of reliable health data creates adverse selection, where buyers cannot distinguish between well-maintained and degraded vehicles, suppressing market value.
Regulatory Action: The findings challenge the validity of current regulatory frameworks (EU, California) that assume reported SOH is meaningful. The authors recommend:
1. Mandatory Standardized Testing: Regular capacity tests under controlled conditions accessible to third parties.
2. SOH Validation Requirements: Manufacturers must demonstrate BMS SOH correlates with independent capacity ( $\rho > 0.7$ ) to certify warranties.
3. Open Data Standards: Requiring access to raw voltage/current data to enable third-party health assessment.
Second-Life Markets: Accurate, independent health assessment is critical for the safe and economic repurposing of EV batteries for stationary storage.

Conclusion: The study concludes that BMS SOH reporting is currently unreliable and unpredictable across manufacturers. A shift toward manufacturer-independent, physics-based health indicators (like $V_{peak}$ and usage-pattern analysis) is necessary to ensure market transparency, fair warranty enforcement, and regulatory compliance.