IEEE 1584-2018 vs 2002 — The Changes That Add Up to Completely Different Results

We ran the same 415 V switchboard through both models. The incident energy result differed by 40%. Same bus gap, same bolted fault current, same clearing time. The only variable was the edition of IEEE 1584 we applied.

That is not a rounding difference. That is the difference between a Category 2 arc flash suit and a Category 4. Between a 15-minute job and a full shutdown. Between a label that says "Danger" and one that says "Warning."

IEEE 1584-2018 is not an incremental update. It is a fundamentally different model, built on a dataset roughly ten times larger than what informed the 2002 edition. If you are still using the 2002 equations — or worse, if your facility labels were calculated under the old model and never rechecked — the numbers on those labels may be wrong in a direction that matters.

The 2002 model treated every arc the same way. It did not distinguish between an arc in open air and an arc inside an enclosure, between vertical electrodes and horizontal ones, or between an arc that terminates on a grounded barrier and one that does not. Every scenario used the same empirical equation.

The 2018 model introduces five distinct electrode configurations:

• VCB — Vertical conductors inside a box (enclosure)
• VCBB — Vertical conductors inside a box, with an insulating barrier at the bottom
• HCB — Horizontal conductors inside a box
• VOA — Vertical conductors in open air
• HOA — Horizontal conductors in open air

Each configuration has its own set of empirical coefficients for arcing current and incident energy. This matters enormously in practice because the arc plasma behaviour differs significantly depending on electrode orientation and the presence of an enclosure. A VCB configuration typically produces higher incident energy at the working distance than a VOA configuration at the same fault level, because the enclosure focuses the arc energy toward the opening where the worker stands.

In the 2002 model, the lack of this distinction meant that some scenarios were overestimated and others were underestimated. The 2018 model corrects both directions.

The 2002 edition treated bus gap as a variable but with limited sensitivity. The empirical model was derived from tests at a narrow range of gap distances, and the equations reflected that narrow base.

The 2018 edition tested gaps from 6.35 mm to 152.4 mm across all five electrode configurations. The result is a model where gap distance genuinely influences the arcing current calculation, not just the incident energy. A wider gap produces a lower arcing current (the arc is harder to sustain), which in turn affects the protective device clearing time and the total energy delivered.

This creates a cascading effect: change the gap, and you change the arcing current, which changes the clearing time on the time-current curve, which changes the incident energy. The 2002 model did not capture this cascade with the same fidelity.

For practical engineering, this means that the actual bus gap in your switchboard — not just a default value from a table — now has a material effect on the result. Measuring it correctly during a site survey is more important than ever.

The 2002 model applied a single enclosure correction factor. The 2018 model introduces a detailed correction based on actual enclosure dimensions — height, width, and depth.

The correction factor accounts for the way arc energy reflects off enclosure walls. A shallow enclosure focuses more energy outward than a deep one. A narrow enclosure concentrates energy more than a wide one. These are not intuitive relationships, and the 2002 model did not capture them at all.

The practical implication is significant for panel builders and facility engineers: the physical dimensions of the switchboard now feed directly into the arc flash calculation. Two switchboards with identical electrical parameters but different physical dimensions will produce different incident energy results. This is physically correct — and the 2018 model finally reflects it.

For engineers performing arc flash studies on existing facilities, this means recording the enclosure dimensions during the site survey, not just the electrical parameters.

Both editions recognised that arcing current is variable — an arc is not a stable impedance. The 2002 edition addressed this with a simple 85% multiplier: calculate at full arcing current, then recalculate at 85% to check if the protective device clears more slowly at the reduced current.

The 2018 edition replaces this with a statistically derived variation model. Instead of an arbitrary 85%, the standard provides equations for the "reduced arcing current" that depend on voltage, electrode configuration, and gap distance. The variation is larger at lower voltages and for certain configurations, which aligns with experimental observations.

This is particularly important at 208 V and 240 V, where the 2002 model's 85% factor was often inadequate. At these lower voltages, arcing current variation can be significantly larger than 15%, and the 2018 model captures this reality. The result: low-voltage arc flash calculations under 2018 often produce higher incident energy than under 2002, because the reduced arcing current pushes the clearing time into slower regions of the protective device curve.

If your facility's arc flash labels were calculated using the 2002 edition and have not been recalculated, there is a reasonable probability that some of them are inaccurate. The direction of the error depends on the specific equipment:

• Low-voltage equipment (208–480 V): The 2018 model generally produces higher incident energy values at low voltages, especially for enclosed equipment. Labels calculated under 2002 may understate the hazard.
• Medium-voltage equipment (1–15 kV): Results vary by configuration. Some MV scenarios produce lower incident energy under 2018 (where the 2002 model was overly conservative), while others remain similar.
• Open-air scenarios: The 2018 model's VOA and HOA configurations sometimes produce lower values than the 2002 model, because the 2002 model applied enclosure-like energy focusing even to open-air scenarios.

The bottom line: you cannot assume the direction of the error without running the calculation. A blanket statement that "2018 gives higher results" or "2018 gives lower results" is incorrect. The answer depends on the specific equipment parameters, and that is precisely why the 2018 model is better — it captures the physical differences that the 2002 model averaged away.

If you are responsible for arc flash studies or PPE selection, here are the practical steps:

1. New studies: Use IEEE 1584-2018 exclusively. There is no technical reason to use the 2002 edition for any new arc flash study.
2. Existing labels: Prioritise recalculation of low-voltage switchboards and any equipment where workers regularly perform energised work. These are the scenarios where the 2018 model most often produces materially different results.
3. Site surveys: Record bus gap distances and enclosure dimensions (height, width, depth) for every piece of equipment. The 2018 model needs this data; the 2002 model did not.
4. Software: Verify that your calculation tool uses the 2018 model, not the 2002 equations with a "2018" label. Ask your vendor which electrode configurations are supported and how enclosure corrections are applied.

An arc flash calculator that implements the full 2018 model — all five electrode configurations, enclosure geometry corrections, and the statistical arcing current variation — gives you results that reflect the physics of the arc. That is what the standard intended.