Multiple arc flash labels on one enclosure: when you need them

Three voltage levels, one steel box, three labels — and a question about whether that is right. The count is right: an arc flash label describes an energized condition, not an enclosure. Voltages that are always live together share one worst-case label. Each level a worker can lock out on its own earns its own label. An arc flash detector adds a second set of numbers, because cutting the clearing to 10 ms from 67 ms leaves about one-seventh of the incident energy at the same fault current. The running example is a three-winding turbine transformer enclosure, the kind of box that raises the question.

How many arc flash labels does one enclosure need?

Count the conditions a worker can meet, not the boxes: a single enclosure needs one label per energized state the study covers, and the worked example below ends at four labels for two voltage levels with and without detector clearing.

NFPA 70E asks for a label on equipment likely to be worked on while energized, reporting the results of the incident energy analysis. The analysis is built from conditions: what is live, at what fault current, cleared in what time. The standard sets no label count per box, so the box count and the condition count are different numbers. One junction box with five circuits at one exposure is one condition. One transformer enclosure with three windings behind separate breakers can be three conditions before the detector states are even added.

In Canada, CSA Z462 carries the same labeling logic, and wind projects there also work to CSA Z468, the wind turbine electrical safety code. The habit is the same under both: define the conditions first, then print one label for each state the door can hide. When one box holds several circuits with different exposures, the same counting decides, as in arc flash incident energy for multiple circuits in one junction box.

When does one worst-case label cover several voltages?

One label covers the lot when a single lockout de-energizes every level and no level can be live alone: print the highest incident energy of the set, with its voltage and boundary, on one label.

This is the usual case for a control circuit. A 120 V control transformer tapped off a 480 V bus goes dead with the bus, so the two voltages form one condition. You put the higher numbers on the label and the whole door is covered. The worst case almost always sits at the highest voltage and the highest available current, because at a fixed clearing time and working distance the incident energy rises with both.

The cost shows up in the clothing. A worker troubleshooting the 120 V circuit wears the 480 V suit, even though the 120 V arc, run through the same formula at its own smaller current, computes to a tenth of the energy or less. If your plant accepts that overhead, one label is the simplest answer to print. If it wants control work in lighter clothing, it needs the per-level study in the next section.

When does each voltage level need its own label?

Each level needs its own label the moment a worker can open the box with that level live and another one locked out: independently lockable levels are separate conditions, and their numbers differ.

The turbine enclosure is built this way. The 690 V secondary and the 400 V auxiliary winding each sit behind their own breaker. A crew can lock out the 690 V side and still find the auxiliary board live, and that state has its own fault current, its own energy and its own PPE.

The payoff is real. In the worked example below, the auxiliary board computes to 0.8 cal/cm² at an 18 in [455 mm] working distance, against 2.9 cal/cm² at the 690 V terminals. A worst-case-only label drags the lower-voltage crew into the heavier suit for nothing. A per-condition label lets each crew wear what its own row calls for, and arc flash PPE for junction boxes: assess the work first keeps that assessment honest. To turn a row into a clothing choice, the category table decides, and a guide to selecting arc-flash PPE by category and incident energy walks it row by row.

Collapse back to one label whenever a level cannot be locked alone. If the auxiliary winding feeds through a shared incomer with the 690 V side, the levels are live together by construction, and the worst case covers the door. Pick the label set to match how your lockouts actually work, and check the single-line before the study runs, not after.

What has to appear on the label itself?

NFPA 70E sets the minimum fields: nominal system voltage, incident energy at the working distance, the arc flash boundary, and at least one of the clothing arc rating or the available incident energy with its working distance.

Those four fields describe exactly one condition, which is why a box with several conditions ends up with several labels or one multi-row plate. Two habits keep the door readable. Print one plate with a row per condition, so the worker compares rows at a glance instead of hunting around the box. And mount the plate outside the door, so the exposure is read before the door swings open, not after.

Most plates also carry the study date and the equipment identification, which matters when you re-run the study: the old plate comes off, the new one goes on, and nothing in between survives. The full detail behind each row, the arcing currents and device settings, stays in the study report. The steel box itself carries no arc rating of its own; what an arc flash rated enclosure actually is draws that line.

How does an arc flash detector change the numbers?

It cuts the clearing time from tens of milliseconds to single digits, and at the same fault current the incident energy falls in the same proportion: 67 ms to 10 ms leaves about one-seventh of the energy.

The linear rule makes the numbers easy to trust. Both the IEEE 1584 empirical model and the Lee formula scale incident energy directly with arcing time at a fixed current and distance. Cut the clearing time to a seventh and the energy follows. A power circuit breaker clears in about 4 cycles, 67 ms at 60 Hz. A typical arc flash relay detects the light and closes its trip output in about 2 ms, the light-to-trip figure relay datasheets quote. The device it trips finishes the job; the worked example assumes 10 ms in total.

That is why the study runs both cases: the detector can be out of service. Testing trips it, a sensor channel fails, an auxiliary supply drops, and the box is back to 67 ms. Each state takes its own label rows. The with-detector numbers also hold only where the sensors watch; a compartment without sensor coverage keeps the without-detector row even with the relay healthy.

What does the label math look like for the turbine enclosure?

Take two live levels, 690 V at 25 kA and 400 V at 12 kA, clear them in 67 ms or 10 ms, and the Lee formula from IEEE 1584 gives four rows from 2.9 cal/cm² down to 0.12 cal/cm² at an 18 in [455 mm] working distance.

Those inputs are worked-example assumptions; replace the fault currents and clearing times with your own study's values before you print anything. The Lee method, the theoretical formula carried in IEEE 1584, reads E = 2,142,000 × V × I × t / D². E comes out in J/cm², with V in kV, I in kA, t in seconds and D in millimetres. Divide by 4.184 for cal/cm², and set D from a boundary energy of 5.0 J/cm², the 1.2 cal/cm² threshold NFPA 70E uses for the arc flash boundary.

For the 690 V row at 67 ms: E = 2,142,000 × 0.69 × 25 × 0.067 / 455² = 2,475,617 / 207,025 = 11.96 J/cm², or 2.86 cal/cm². The boundary sits at √(2,475,617 / 5.0) = 704 mm. The 10 ms row divides the time by 6.7, so the energy falls by the same 6.7 to 0.43 cal/cm², and the boundary shrinks by √6.7 to 272 mm. At the auxiliary level and 12 kA the same arithmetic gives 0.8 cal/cm² at 67 ms and 0.12 cal/cm² at 10 ms, with boundaries of 371 mm and 143 mm.

The decision follows the table. Two independently lockable levels times two detector states give four rows, printed as four rows on one plate. If the 690 V and auxiliary sides are always live together and the relay is always in service, one row, the 2.9 cal/cm² worst case, covers the door. The box geometry sits behind these numbers too: IEEE 1584 corrects for enclosure size, and why shallow enclosures calculate lower incident energy shows that effect. The medium-voltage winding, typically 34.5 kV on a turbine pad, takes its own row at its own working distance; 910 mm [36 in] is the usual medium-voltage value in the IEEE 1584 method.

What do you order around the label set?

One enclosure order covers it: spec a 316L stainless box for the salt-air turbine site, a flat exterior door face sized for the full condition plate, and gasketed Type 4X construction for the washdown environment.

316L earns its place at coastal and offshore turbine sites, where salt spray pits 304 in crevices and under gaskets; 304 holds indoor and sheltered outdoor service. Type 4X means hose-directed water plus corrosion resistance, the right call for tower-base washdown, and the type definitions sit in the NEMA enclosure types guide. Because our boxes carry manufacturer ratings, write the spec as rated NEMA 4X / IP66 by the manufacturer and not UL listed. When the project insists on a UL label, the UL 50E listing guide explains how listing to UL 50E works and when an inspector asks for it.

Size the door face with the plate in mind. Four rows of voltage, incident energy and boundary print legibly at 10 pt inside a clear patch about 100 by 65 mm [4 by 2.5 in]. Engraved stainless or UV-stable polyester labels hold up outdoors, where paper laminates lift after a season or two. You can spec the material, the door face and every cutout in one pass through custom enclosures, and the stainless size finder filters wall-mount boxes by grade and size. The label set rides on the drawing approval, so the plate plan reaches the shop before the box is cut, and the finished box arrives in 10 business days from that approval.

Catalog dimensions to compare

Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.

Questions