Arc Flash Incident Energy: How the 2018 IEEE 1584 Revision Changed Everything

The original IEEE 1584-2002 model was based on approximately 300 test data points covering a voltage range of 208 V to 15 kV. It used two separate equation sets: one for systems below 1 kV and one for systems from 1–15 kV. The equations calculated arcing current and incident energy using a relatively small number of input variables:

• System voltage (V)
• Bolted fault current (kA)
• Working distance (mm)
• Arc duration (seconds)
• Enclosure type (open air, box, or cubic box for MV)
• Bus gap (fixed values: 25 mm for panelboards, 32 mm for LV switchgear, 104 mm for 5 kV, 152 mm for 15 kV)

The model had several known limitations that practitioners had documented over years of application:

1. Fixed bus gaps. The 2002 model used predetermined bus gap values for each equipment category. In reality, bus gaps vary significantly between manufacturers and equipment types. A custom-built mining switchboard may have 50 mm bus gaps rather than the assumed 32 mm.
2. No electrode configuration. The model did not distinguish between vertical and horizontal bus arrangements or whether the arc would be directed toward or away from the worker. The arc behavior differs profoundly between these configurations.
3. Fixed enclosure size. The “box” enclosure correction was based on a single standard box size (508 mm × 508 mm × 508 mm for LV). Real enclosures range from small distribution panelboards to large switchgear lineups.
4. Limited test data. With approximately 300 data points, the statistical confidence of the model was limited, particularly at the boundaries of its valid range.
5. Voltage range. The model was validated only from 208 V to 15 kV. Systems outside this range required alternative methods (Ralph Lee equations for higher voltages).

The IEEE 1584-2018 revision represents a fundamental reconstruction of the arc flash calculation methodology. The key changes are:

1. Expanded test database: The model is based on 1,800+ test data points — six times the original dataset. Tests were conducted at multiple laboratories worldwide using standardized test setups.

2. Five electrode configurations:

The electrode configuration is the single most impactful new variable. The VCBB configuration, where the arc is directed toward the worker by a barrier at the bottom of the enclosure, consistently produces the highest incident energy. The VOA configuration (open air, vertical) typically produces the lowest.

3. Variable bus gap: Rather than fixed values, the engineer inputs the actual bus gap in millimeters. The valid range is 6.35 mm to 76.2 mm for LV systems and up to 254 mm for MV.

4. Variable enclosure dimensions: Width, height, and depth of the enclosure are input parameters that affect the enclosure correction factor. Larger enclosures produce less focusing of arc energy; smaller enclosures concentrate it.

5. Expanded voltage range: The model is validated from 208 V to 15 kV as a continuous range, eliminating the discontinuity between the LV and MV equation sets of the 2002 model.

6. Multi-step calculation process: The 2018 model uses intermediate calculations for arcing current (with variation factor), normalised incident energy, arc size correction, and enclosure correction. The calculation is more involved but each step has physical meaning.

The following table shows incident energy calculated under both models for a standardized 480 V switchboard scenario: 30 kA prospective fault current, 32 mm bus gap, 508 mm × 508 mm × 508 mm enclosure, 457 mm working distance, 200 ms clearing time.

The results span a 4.7× range (4.2 to 19.8 cal/cm²) depending on electrode configuration. The 2002 model’s single result of 8.6 cal/cm² falls roughly in the middle — but it could be either dangerously non-conservative (for VCBB) or excessively conservative (for HCB).

Additional scenarios showing the divergence:

Across all scenarios, the VCBB configuration consistently produces results 89–142% higher than the 2002 model. The VCB and HCB configurations generally produce lower results (17–45% lower). This means the 2002 model was systematically non-conservative for equipment with VCBB geometry — which includes a significant fraction of LV switchboards where bus bars terminate at a backplane barrier.

The electrode configuration selection is arguably the most critical decision in a 2018-edition arc flash study. IEEE 1584-2018, Clause 4.7 and Table 1 provides guidance:

The challenge for existing facilities is that determining the electrode configuration requires physical inspection of the equipment interior, not just a review of single-line diagrams. A desktop arc flash study cannot reliably assign electrode configurations without field verification. For many facilities with hundreds of panels and switchboards, this inspection campaign represents a significant effort that was not required under the 2002 edition.

IEEE 1584-2018, Clause 4.4 introduces a mandatory calculation at reduced arcing current. The standard provides a variation factor (VarCf) that is applied to the calculated arcing current to determine a reduced value. The engineer must calculate incident energy at both the full arcing current and the reduced arcing current, and use the higher result.

The rationale is that arc flash is a stochastic event — the arcing current varies from event to event depending on arc initiation conditions, electrode surface state, and environmental factors. A lower arcing current may result in a longer clearing time if the protective device is operating in its time-delayed region (inverse time characteristic), producing higher incident energy despite lower instantaneous arcing power.

Example impact of reduced arcing current:

In the second scenario, the reduced arcing current of 10.8 kA falls below the MCCB’s instantaneous trip threshold, causing the breaker to operate on its long-time delay curve at 850 ms rather than instantaneously at 30 ms. The incident energy more than doubles compared to the full arcing current case. This scenario — where reduced arcing current produces the governing result — is more common than many engineers expect, particularly for systems with high prospective fault currents and protection devices with high instantaneous trip settings.

Every facility with an arc flash study performed under IEEE 1584-2002 faces a decision: when (not whether) to update the study to the 2018 edition. The factors that should trigger an immediate re-study are:

1. Equipment with VCBB configuration. If any switchboards or MCCs have vertical bus bars terminating at a barrier (which directs arc energy toward the worker), the 2002 study is likely non-conservative by 80–140%. These panels should be re-studied immediately.
2. Equipment with high prospective fault current (>30 kA at LV). The divergence between the 2002 and 2018 models increases with fault current. High-fault-current installations have the most to gain from re-study.
3. Protection devices operating in the long-time delay region at arcing current. The 2018 reduced arcing current requirement may reveal scenarios where clearing time is dramatically longer than assumed, producing governing incident energies not captured by the 2002 study.
4. Any incident or near-miss. An arc flash event at the facility is an unambiguous signal to update all arc flash labels to the current standard.
5. Regulatory audit or compliance renewal. Many jurisdictions now reference the 2018 edition. Continuing to label equipment based on the 2002 edition may not meet current regulatory expectations.

For facilities where an immediate re-study is not feasible, a screening approach can prioritize equipment: apply the 2018 VCBB configuration to the highest-fault-current panels as a desktop check. If the results are within the same PPE category as the existing labels, the urgency is lower. If the category increases, those panels should be re-studied and relabelled as a priority.

Standards referenced: IEEE 1584-2002, IEEE 1584-2018 (Clauses 4.1–4.9, Annex D), NFPA 70E:2024 (Article 130), IEC 60909-0:2016.