Arc flash and electrical safety — the label, the study behind it, and the work practice

An arc flash is the sudden release of the energy available at a point in the electrical system, through air, as heat, light, pressure and vaporised metal. It does not need a person to make a mistake with a conductor — a dropped spanner, a failed insulator, a rodent, or a termination that has been deteriorating for a year will do it.

What makes it an integrity subject rather than only a safety one is that the severity is set by decisions made elsewhere: by the size of the supply, by the protection settings, and by whether anyone has checked that the study describing those things is still true.

What incident energy depends on

Incident energy is the thermal energy arriving on a surface at a given distance from the arc, quoted in calories per square centimetre. The number that matters in practice is 1.2 cal/cm2, roughly the onset of a second degree burn on bare skin, because that is the value that defines the arc flash boundary.

Three quantities set it.

The available fault current. Fixed by the supply and the impedance between it and the board. Not within the plant’s gift, and in any case the relationship is not simple — a lower fault current can produce a higher incident energy if it makes the upstream relay slower.

The arcing duration. This is the clearing time of the upstream protective device, taken from the grading study. Energy is roughly proportional to it, so halving the clearing time roughly halves the energy.

The working distance. How far the worker’s face and chest are from the arc. Energy falls off quickly with distance, which is why the same board is rated differently for rack-out work and for meter reading through a window.

IEEE 1584, and what a calculation is worth

IEEE 1584 gives the empirical model that almost every study uses: arcing current from bolted fault current, incident energy from arcing current, duration, enclosure geometry and working distance. The 2018 revision widened the test basis and changed results on many systems, which is in itself a reminder of what kind of document it is.

It is a model fitted to test data, and it carries the limits of that data — defined voltage and current ranges, defined electrode configurations, defined enclosure sizes. A real cubicle with a cable box full of terminations is not one of the tested geometries, and the honest position is that the study gives a well-founded estimate rather than a measurement.

That does not make it optional. It makes the inputs matter more than the software. A study built on a single line diagram that does not match the plant, or on relay settings read from a stale study rather than from the relays, produces confident numbers about a system that does not exist.

The work practice, which is where lives are actually saved

The hierarchy is not complicated, and the order is the point.

1. Work dead. The great majority of live working on an industrial plant is live because of scheduling convenience, not because the task requires it.
2. Isolate and secure. Open the isolator, lock it with a personal lock on a multi-hasp, and hold the key. One lock per person at risk.
3. Prove dead. Prove the tester on a known live source, test at the point of work on every conductor pair, then prove the tester again. An isolator position indicator is a claim, not a test.
4. Earth where required. On HV, applied earths are what protect against a back-feed or a switching error.
5. Where live work is unavoidable, it needs a specific risk assessment, a second person, and clothing rated above the incident energy at that working distance.

Clothing, and the point where it stops helping

Arc-rated clothing is specified by its arc thermal performance value, which has to exceed the calculated incident energy at the working distance. Three practical points matter more than the specification.

Non arc-rated synthetic clothing underneath will melt onto skin, so the whole clothing system counts, not the outer layer. Hands and face are the parts most often injured and the parts most often left out. And above roughly 40 cal/cm2 the pressure wave and the sound are injuring in their own right, so a very high rating is an instruction to eliminate the task rather than a solution.

The Indian statutory overlay

On an Indian plant the legal requirements come from the Electricity Act and the CEA Measures Relating to Safety and Electric Supply Regulations, enforced by the state electrical inspectorate, with the Factories Act and state factory rules alongside them. These set authorisation and competency of persons working on electrical installations, clearances, earthing, and the requirement for the installation to be maintained in a safe condition.

What they do not do is prescribe an incident energy calculation or a clothing rating. So the usual position on a well-run site is that the CEA Regulations are what the plant is held to, and IEEE 1584 and NFPA 70E are the methods adopted to satisfy them properly. That distinction is worth being clear about, because a contractor’s NFPA-based procedure does not by itself discharge a statutory duty, and a statutory inspection does not by itself confirm the arc flash study is current.

Where this goes next

• Protection coordination and grading — the clearing time that the energy calculation depends on.
• Switchgear, transformers and protection testing — proving the device actually operates in the time the study assumed.
• Electrical testing and diagnostics — finding the deteriorating termination before it starts the arc.
• Distribution and the single line diagram — the document every study is built on.