Skip to content
LogoBATRIS
Second-life battery assessment

Know what a used
battery is worth

A capacity test costs a day on a bench cycler. This platform estimates state of health from an ordinary charge cycle, tells you why the pack is degrading, checks it against its own safety limits, and issues a signed passport a buyer can verify.

Estimated state of health
78.4%

90% interval 74.9–81.9% · 1.57 Ah remaining

Second-life grade
B
Stationary storage
Passport
Signature valid
Ed25519 · issuer key
100%90%80%70%60%80% end of first lifeCycles
Estimated SOH90% confidence bandEnd of first life
2.67SOH pts
Best measured error
4cells
Leave-one-out validation
6numbers
Minimum useful input
What it reports

Six answers from
one charge cycle

Each block below is produced by a separately validated component, and each is labelled with how it was obtained. A number the model inferred is never presented the same way as a number that was measured.

Estimated

State of health

A capacity estimate with a calibrated 90% interval, drawn from how the charge divides between its constant-current and constant-voltage phases.

Explained

Degradation attribution

Which mechanism is taking the capacity — lithium inventory loss, resistance growth, thermal stress — and the measured signals behind each claim.

Derived

Safety envelope

A risk score with named drivers, plus derated charge current, SOC window and temperature limits computed from the pack's own format specification.

Detected

Anomaly detection

Physical rule checks, statistical outlier detection and trajectory deviation, each reported separately so silence from a detector is distinguishable from a clean result.

Graded

Second-life grade

A reuse grade on the point estimate, with the grade marked ambiguous whenever the confidence interval straddles a boundary.

Verifiable

Signed passport

An Ed25519-signed document carrying the assessment, the model provenance and the method labels. Alter one field and verification fails.

How it works

Four steps, in this order

The order matters, which is why these are numbered: each step determines which model the next one is allowed to use.

  1. 01

    Charge it

    Run one ordinary charge. Either log it as CSV, or read six numbers off the charger display when it finishes.

  2. 02

    Submit what you have

    There is no fixed input set. The platform picks the richest tier your data supports and uses the model trained for exactly that tier.

  3. 03

    Read the estimate

    Health, safety, anomalies and reuse grade, each with the accuracy measured for your tier — not a headline number borrowed from the best case.

  4. 04

    Issue the passport

    Sign the assessment into a portable document. A buyer verifies it against the issuer's public key without trusting you.

Measured accuracy

The number you get is the one measured for your inputs

Mean absolute error in SOH percentage points, measured by leave-one-battery-out cross-validation on four NASA cells. Each tier has its own separately trained and separately validated model, so a missing input never becomes a silent null in a model that never saw one.

TierWhat you supplyMAER²Output
Tier 1
Telemetry + impedance
Full charge curve plus EIS values2.67 pts0.867Full report
Tier 2
Telemetry
Full charge curve, no impedance2.75 pts0.858Full report
Tier 3
Charge summary
Six hand-typed numbers2.86 pts0.845Full report
Tier 4
Minimal
Charge phase split only5.58 pts0.127Indicative
Tamper-evident by construction

A claim a buyer can check without trusting the seller

The passport is a JSON document signed with an Ed25519 key. It carries the health estimate, the confidence interval, the model provenance hash and an explicit label saying the figure was estimated rather than measured. Change the SOH from 78% to 97% and the signature stops verifying — which you can demonstrate from the dashboard in one click.

  • Ed25519 detached signature over a canonical serialisation
  • Method labels travel with the numbers, so an estimate can never be read as a test result
  • Model version, training-data hash and validation method embedded in the payload
  • Verify against a published key, or paste the document into any Ed25519 verifier
  • A QR code provides instant access to the battery passport for quick verification and lifecycle tracking
passport.jsonverified
{
  "payload": {
    "battery": { "battery_id": "B0005" },
    "health_estimate": {
      "soh_percent": 78.4,
      "confidence_interval_90": [0.749, 0.819],
      "method": "ESTIMATED"
    },
    "second_life_assessment": {
      "grade": "B",
      "grade_confidence": "MEDIUM"
    },
    "certified_test_status": "NOT_PERFORMED"
  },
  "signature": {
    "algorithm": "Ed25519",
    "value": "9f2c…a71b"
  }
}

Change soh_percent to 97.5 and the signature stops verifying. You can do exactly that from the dashboard.

From assessment to record

The assessment doesn't end with a number

BATRIS turns the result of a battery assessment into a portable, verifiable record — one that can be shared, checked and carried forward with the battery.

The passport brings the battery’s health estimate, confidence interval, safety findings, degradation signals and second-life assessment into a single machine-readable record.

Point it at a battery

Four NASA cells are loaded and ready to assess. Or bring your own charge log and see what your level of information actually buys you.