Practical Battery Arc Flash Models

Overview

The study asks whether arc-flash incident energy in battery systems can be calculated both accurately and practically. NFPA 70E Annex D has referenced the maximum power transfer method since 2012; it tends to overestimate incident energy and has led to over-prescription of PPE. The proposed method predicts self-extinguishing time from battery voltage, short-circuit current, arc gap and electrode configuration (inline, open parallel, closed parallel), modelling electrode erosion and treating arc length as a random variable.

The method was tested against 228 published battery arc-flash tests and compared with the NFPA 70E Annex D maximum power transfer method and an arc-resistance method. Mean overestimation was 11.88 cal/cm² for maximum power transfer and 7.38 cal/cm² for the arc-resistance method, against 2.79 cal/cm² for the proposed ≤5% exceedance (EaR) self-extinguishing method. The maximum power method predicted a higher PPE category than measured in 90% of tests; the ≤5% EaR method selected the correct minimum PPE category in 67% of tests and underestimated in less than 1%.

In inline configurations electrode erosion dominates self-extinguishing; in parallel configurations Lorentz forces dominate. Terminating parallel conductors at an insulating boundary such as a breaker can greatly extend arc duration and make it less predictable, and in practice DC load overcurrent protection usually limits arc duration more than self-extinguishing does. Sensitivity: for a 520 V, 20 kA, 12.7 mm gap case, a 1% increase in the tuning parameter β raised the median inline arc time by 12.97% (1.19 s to 1.34 s).

Limits: the method is valid below 600 V for inline and closed parallel configurations and below 1 kV for open parallel; for short-circuit currents below 10 kA (open parallel) or 20 kA (inline and closed parallel); and only for batteries at rest on low-inductance circuits. An arc-in-a-box coefficient of 1.52, calculated for panelboards, was used for enclosed experiments. Outside these ranges, for example on 800 VDC racks, it should not be used directly.

Key points

1. Across 228 published tests, mean overestimation was 11.88 cal/cm² for the maximum power transfer method and 7.38 cal/cm² for the arc-resistance method, versus 2.79 cal/cm² for the proposed ≤5% exceedance (EaR) self-extinguishing method.
2. The maximum power method predicted a higher PPE category than measured in 90% of tests; the ≤5% EaR method selected the correct minimum PPE category in 67% of tests and underestimated in less than 1%.
3. An arc-in-a-box coefficient of 1.52, based on the value calculated for panelboards, was used for enclosed-space experiments.
4. Validity limits: below 600 V for inline and closed parallel and below 1 kV for open parallel; short-circuit current below 10 kA (open parallel) or 20 kA (inline and closed parallel); only batteries at rest on low-inductance circuits.
5. Terminating parallel conductors at an insulating boundary such as a breaker can greatly extend arc duration and make it less predictable; in practice DC load overcurrent protection usually limits arc duration more than self-extinguishing does.
6. For a 520 V, 20 kA, 12.7 mm gap test case, a 1% increase in the tuning parameter β raised the median inline arc time by 12.97% (1.19 s to 1.34 s).

Using it in design

• In battery-room arc-flash labelling, calculate the self-extinguishing model alongside the maximum power method rather than using the latter as the only reference, and check cases where excessive PPE categories make maintenance impractical.
• Take inputs (voltage, short-circuit current, gap, electrode configuration) from the battery data sheet and the physical layout, and check the validity ranges (600 V, 1 kV, 10-20 kA) at the start of the calculation.
• Where parallel conductors terminate at an insulating boundary, assess arc duration together with protective device clearing time, and treat DC overcurrent protection as the duration limiter.

Who should read it

Electrical safety engineers who perform arc-flash studies and produce labels for battery systems and BESS.

Document contents

1. I. Introduction
2. II. Arc Flash Physics
3. III. Experimental Data (A. Inline configurations, open parallel and closed parallel configurations)
4. IV. Methods (A. Maximum power transfer method, B. Arc resistance method, C. Erosion model, D. Arc length as a random variable, E. Arc elongation in parallel configurations)
5. V. Results
6. VI. Discussion on Selecting a Configuration
7. VII. Conclusion

Related standards

• NFPA 70E Annex D

Read the document

Publication details

Authors

David Rosewater; Lloyd B. Gordon; William Cantor

Publisher

IEEE Transactions on Industry Applications (Institute of Electrical and Electronics Engineers (IEEE))

Year

2026

Volume

62

Issue

3

Pages

5512-5521

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.1109/TIA.2025.3637673

Licence

CC BY 4.0 — share with attribution

Licence evidence

ieeexplore.ieee.org

Citation