Skip to content

Thermal propagation

Verifies that thermal runaway in a single cell does not propagate to a pack-level fire or explosion, that an alarm reaches the driver in time, and that smoke does not enter the cabin. The headline test of GB 38031-2025.
Clause (method) 8.2.7.2 (refers to Appendix C)
Clause (pass criteria) 5.2.7 b)
Object pack / system / entire vehicle
Status vs. 2020 revised (both requirements and method)
Observation period Until all monitoring points ≤ 60 °C, minimum 2 h (C.5.3.8)

For the full procedural detail (the OEM documentation package, sensor placement rules, trigger-method specifications, and the runaway-confirmation logic), see the Thermal propagation deep dive.

Pass criteria

"The battery pack or system or the entire vehicle, after undergoing the thermal propagation analysis and verification as per 8.2.7.2, should: 1) No fire, no explosion; 2) Provide a thermal event warning signal, and the warning signal should be sent no later than 5 minutes after thermal runaway of the battery cell is triggered; 3) For battery pack or system-level testing, provide technical documentation indicating that smoke does not pose a danger to the passenger compartment within 5 minutes before and after the thermal event warning signal is issued; for vehicle-level testing, smoke should not enter the passenger compartment within 5 minutes before and after the thermal event warning signal is issued."

Source: GB 38031-2025, clause 5.2.7 b) (PDF p. 12).

Engineering interpretation (non-normative)

Decision tree for choosing the test object and SOC.

flowchart TD
    Start([Choose test object]) --> Q1{Externally chargeable<br/>battery?}
    Q1 -->|yes| SOC95[Adjust SOC ≥ 95 %<br/>of mfr's max working state]
    Q1 -->|no — vehicle-energy only| SOC90[Adjust SOC ≥ 90 %<br/>of mfr's max working state]
    SOC95 --> Lev{Test level}
    SOC90 --> Lev
    Lev -->|pack| Pack[Run on pack +<br/>provide smoke-safety doc]
    Lev -->|system| Sys[Run on system +<br/>provide smoke-safety doc]
    Lev -->|whole vehicle| Veh[Run on vehicle —<br/>smoke must NOT enter cabin]

Pre-conditions

  • Sample: A battery pack, system, or entire vehicle. (C.5.1)
  • Indoor environment: ambient > 0 °C, RH 10–90 %, atmospheric pressure 86–106 kPa or wind speed ≤ 2.5 km/h. Pack/system pre-test temperature: 22 °C ± 5 °C. (C.5.2 a)
  • SOC: ≥ 95 % of mfr-specified maximum for externally chargeable packs/systems; ≥ 90 % for packs charged solely via the vehicle. (C.5.2 b)
  • Modifications: Avoid unnecessary modifications. The manufacturer must submit a list of modifications made to the test sample. Confirm SOC after modification. (C.5.2 c)
  • System state at start: Pack/system follows the control strategy used during the 1-hour power-down state. For vehicle-level: dashboard and central screen normal, no warnings, in-cabin video capture set up, vehicle in park, doors/windows/sunroof closed, A/C off. (C.5.2 d)
  • Excluded chemistries: Nickel-hydride packs/systems are excluded from clause 5.2.7.

Test parameters

Parameter Value Source
Trigger object A cell near the centre of the pack, surrounded by other cells C.5.3.2
Trigger method Needle penetration or external heating or internal heating plate (recommended; another method is permitted with justification) C.5.3.1
Voltage / temperature sampling interval < 1 s C.5.3.6 a)
TR confirmation rule (a OR b) AND c: a) voltage drop > 25 %, b) temp at monitor reaches mfr max operating T, c) dT/dt ≥ 1 °C/s for > 3 s C.5.3.7
Warning signal latency Issued ≤ 5 min after TR is triggered 5.2.7 b) 2)
Smoke window (pack/system) Documentation: smoke does not endanger passenger compartment in 5 min before / 5 min after warning 5.2.7 b) 3); C.3.5 e)
Smoke window (vehicle) Smoke must not enter passenger compartment in 5 min before / 5 min after warning 5.2.7 b) 3)
Outcome No fire, no explosion 5.2.7 b) 1)
Observation end All monitoring points ≤ 60 °C, minimum 2 h C.5.3.8

For trigger-method-specific parameters (needle dimensions, heating power tables, internal heating plate dimensions), see trigger methods.

Procedure

Steps from Appendix C. The OEM-side analysis-and-verification report is the framing; the verification test (C.5) is the on-bench part.

  1. The manufacturer prepares the three-report package per Table C.1: alarm signal description (C.2), pack/system safety documentation (C.3), and the test report (C.4 — issued by the test agency). See the OEM documentation package.
  2. Confirm test conditions: environment, SOC, sample state, vehicle state if vehicle-level. (C.5.2)
  3. Choose the trigger cell — near the pack centre, surrounded by other cells. (C.5.3.2)
  4. Choose a trigger method — needle, external heating, or internal heating plate. (C.5.3.1, C.5.3.3–C.5.3.5)
  5. Lay out monitoring points per C.5.3.6 placement rules: voltage and temperature sampling interval < 1 s.
  6. Apply the trigger and monitor. Stop the trigger when TR is confirmed per the (a OR b) AND c rule (C.5.3.7), or — for the heating methods — when the monitor temperature reaches 300 °C without TR. (C.5.3.4, C.5.3.5.2)
  7. Record the time of the thermal event warning signal. It must be ≤ 5 min after TR was triggered. (5.2.7 b) 2)
  8. Record fire/explosion events (none allowed) and document smoke trajectory in the warning ± 5 min window. For vehicle-level, in-cabin video must show no smoke ingress. (5.2.7 b) 1) and 3)
  9. After triggering, observe the test object at the test environment temperature until all monitoring points are ≤ 60 °C, with a minimum 2 h observation. (C.5.3.8)
  10. If the trigger method does not produce TR, the manufacturer must demonstrate that none of the three recommended methods would cause TR in this cell. (C.5.3.7, last paragraph)

After-test observation

Observe the test object at the test environment temperature until temperatures at all monitoring points are ≤ 60 °C. Minimum observation time: 2 h. (C.5.3.8)

In-test monitoring requirements: voltage and temperature sampling interval < 1 s (C.5.3.6 a); video capture of the cabin if vehicle-level (C.5.2 d); recording of the warning signal timestamp.

What changed from GB 38031-2020

  • Pass requirement re-anchored to "no fire, no explosion". The 2020 standard required only a 5-minute warning before any thermal event reached the cabin; the 2025 standard requires no fire and no explosion at the pack/system or vehicle level and the 5-minute warning and a smoke-safety condition.
  • Appendix C is now normative, not informative. The three-report structure (Table C.1), the three recommended trigger methods, the (a OR b) AND c TR-confirmation rule, and the < 1 s sampling interval are all explicit.
  • Smoke condition added for the warning ± 5 min window (5.2.7 b) 3) and C.3.5 e)).

Migration impact: New type approvals from 2026-07-01. Already-approved models from 2027-08-01. See Re-certification timeline. This is the most consequential test for re-certification — packs designed against the 2020 standard may need additional propagation barriers, vent paths, or cell-isolation features to meet the 2025 "no fire" bar.

Engineering notes (non-normative)

The notes below are practical interpretation, not part of the standard.

Engineering note (non-normative): The trigger methods in C.5.3.3–C.5.3.5 are recommended, not mandatory (C.5.3.1). A manufacturer can propose another method, but if that method fails to trigger TR, the burden falls back to demonstrating that none of the three recommended methods would have triggered TR either — a much harder demonstration than just running the test once. Choose the recommended method most likely to actually trigger.

Engineering note (non-normative): The 5-minute warning clock starts at TR trigger, not at TR confirmation. In practice the alarm system needs to detect the precursor signals (voltage drop, dT/dt) and emit the warning before the cell fully runs away. Building in the alarm threshold tighter than the (a OR b) AND c confirmation logic is a good idea.

Engineering note (non-normative): The smoke clause is enforced two different ways. At pack/system level it is a documentation requirement (C.3.5 e) — show that smoke would not endanger the cabin, typically by reasoning about pack vent paths and the as-installed orientation. At vehicle level it is an observed fact: in-cabin video over the 10-minute window must show no smoke ingress. Vehicle-level testing is the stricter case.