EV Battery Pack 8D — Thermal Runaway Root Cause & Preventive Design

Problem Overview

**Product**: 400V / 60kWh EV battery pack with 96s3p configuration using NMC pouch cells

**Failure Mode**: Battery pack shut down during DC fast charging at an ambient temperature of 38°C. BMS reported cell module #7 over-temperature (65°C internal estimated). Post-event inspection showed 3 pouch cells in module #7 with 10-15% swelling.

**Detection**: BMS over-temperature protection triggered pack disconnect at 62°C. No thermal runaway occurred — the protection system functioned correctly.

**Customer Impact**: Vehicle disabled during road trip. Fast charging unavailable until root cause resolved.

D2 Problem Description (5W2H)

**What**: 3 cells in module #7 swollen 10-15% with capacity degradation to 78% SOH (other modules at 97-99% SOH)

**Where**: Module #7 (center-rear position in pack, lowest airflow zone)

**When**: During 2C DC fast charge at 38°C ambient, after 14 months and 45,000 km of service

**How Many**: 3 cells affected. Module #7 total cell count: 12 cells in 3p4s configuration

**How Detected**: BMS cell voltage deviation + internal temperature estimate exceeded threshold

**Severity**: Vehicle immobilized, no safety incident, warranty claim

D4 Root Cause Analysis

5-Why Chain

1. Why did cells swell? → Internal gas generation — GC-MS confirmed electrolyte decomposition products (CO₂, ethylene, propylene)

2. Why electrolyte decomposition? → Module #7 operating temperature averaged 8°C higher than pack average over vehicle life

3. Why higher temperature? → Module #7 located at center-rear position with lowest cooling airflow — CFD analysis showed 40% lower flow rate vs. pack average

4. Why did thermal management not compensate? → BMS temperature control used pack-average coolant temperature, not per-module cell temperatures

5. Why per-module compensation not designed in? → Thermal system design assumed uniform airflow — no CFD validation of production-intent pack enclosure

**TRC**: Localized overheating in module #7 due to non-uniform cooling airflow, causing accelerated electrolyte decomposition

**MRC**: Thermal management system designed with uniform airflow assumption — no per-module temperature feedback control

Additional Findings

Module #7 cells were from a different production lot than the other 11 modules

Cell tab welding in module #7 showed 12% higher internal resistance → additional I²R heating

BMS temperature estimation algorithm used cell surface sensors (NTC thermistors) which lag internal cell temperature by 3-5°C during fast charging

D3 Containment

Push OTA BMS update: limit DC fast charge rate to 1C when pack average temperature exceeds 35°C

Replace module #7 in affected vehicle under warranty

Inspect 50 vehicles in same production batch: thermal imaging scan during fast charge to identify cooling imbalance

Issue service bulletin: recommend dealer inspection of battery pack cooling ducts for debris or blockage

D5 Permanent Corrective Action

1. **Cooling System Redesign**: Add airflow guide vanes to equalize flow distribution across all modules. CFD-validated design showing <3°C inter-module temperature difference

2. **BMS Firmware Update**: Per-module temperature control — derate individual modules based on their actual cell temperature, not pack average

3. **Cell Internal Temperature Model**: Replace surface NTC-only estimation with Kalman filter-based internal temperature model using impedance tracking

4. **Cell Tab Welding**: Add resistance measurement as in-process quality gate. Reject modules with >5% inter-cell resistance variation

D7 Prevention

Update thermal system design review: require CFD analysis of production-intent enclosure before design freeze

Deploy thermal imaging audit on production line: 100% of battery packs scanned during first charge cycle

Cloud-based battery analytics: flag vehicles with module-level temperature imbalance >5°C for proactive service

DFMEA update: increase Occurrence rating for cooling imbalance from 3 to 6

Supplier quality: add cell internal resistance screening to incoming inspection

Key Takeaways

1. Pack-level uniform cooling assumptions are dangerous — CFD validation must use production-intent geometry

2. Per-module BMS temperature control is critical when cells have production variation in internal resistance

3. Cell surface temperature sensors lag internal temperature during fast charge — internal temperature estimation is needed

4. OTA updates enable containment without physical recall — a key advantage of connected EVs

5. Cloud-based fleet analytics can identify systemic thermal issues before individual vehicles fail