BattScan-020V performed EIS and gathered AC Impedance data on the battery modules. The BattScan-020V was able to predict battery capacity:

● within 98% accuracy in a 5 minute test,

● within 96.5% accuracy in a 30 second test, and

● within 97% accuracy after only 50 battery samples.

Test Procedures and Benchmarking

These results were achieved benchmarking against a 6 hour cycling test. The UL 1974-compliant test procedure includes:

1. Battery pack disassembly and physical inspection

2. Open circuit voltage check

3. Module-level Electrochemical Impedance Spectroscopy (EIS) at uncontrolled warehouse temperature (18°C ±7°C)

4. CC-CV charge at 1/3 C-rate CC, 1A cutoff CV uncontrolled warehouse temperature (18°C ±7°C)

5. CC discharge at 1/3 C-rate at uncontrolled warehouse temperature (18°C ±7°C)

6. UL 1974 DC Internal Resistance check, with between 1/2 C-rate to 1 C-rate pulse at ~80% and 20% SOC

7. Module-level EIS at controlled temperature (25°C ±3°C)

Model Development

Of the over 500 used battery modules, a handful were rejected due to physical defects. Of the remaining 400+ battery modules, about 80% were used as a training set, and the remaining 20% were used as the test set. The test set represents “real-world” samples - batteries that the model has never seen before. The purpose is to validate how well the model behaves with new data. The results, shown below, suggest that the trained model can predict battery module capacity within a minutes-long test with a high degree of confidence.