Battery Cell Analytics for RUL and Warranty Risk
Battery Cell Analytics: Predict Capacity Fade and RUL with Dataraft – NASA Dataset Case Study
EV warranties, grid storage, and aerospace batteries live or die by cell health prediction. Capacity fade isn’t uniform—temperature, voltage profiles, and charge rates accelerate it unpredictably.
Dataraft’s Cell Analytics module uses NASA’s Li-ion battery dataset (4 cells: B0005/6/7, B0018) to answer real-world questions: Which cell fails first? What drives fade? Cycles to 50% capacity?
Dataset & Operational Profiles
Cells run through CC-CV charging (1.5A to 4.2V, hold to 20mA) and CC discharge (2A to 2.7–2.5V cutoffs), with EIS impedance sweeps. EOL: 30% fade (2Ah → 1.4Ah).
Dataraft configured studies with capacity fade as target; predictors: cycle number, avg voltage/charge current, max temp. Domain rules flagged anomalies (e.g., voltage deviations).
Single-Cell Deep Dive (B0005)
Exploration: All charges hit 4V; discharge capacity fades gradually, accelerating post-cycle 35.
Heatmap: Avg discharge voltage strongest fade driver; max temp secondary.
Model: Quantifies cycles to 50% capacity; avg voltage most influential.
Multi-Cell Comparison
All cells charge correctly; B0006 discharges fastest.
Heatmap: Max temp impacts fade; avg charge current drives voltage.
Model: Predicts drain to 70% capacity in X cycles for B0006.
Impact for Battery Teams
Warranty Risk: Proactive RUL flags reduce field failures.
Design Iteration: Voltage/temp as levers for better cells.
Scalability: Handles NASA-scale datasets; APIs for production monitoring.