Seal-Off Fittings: Where the Flame Path Has to Stop

Forty-one fittings, and the fourth one was hollow

The commissioning walk at a solvent day-tank building came apart over a screwdriver. Forty-one sealing fittings on the electrical drawing, every one installed, threaded up hard, painted the same grey as the rack they were bolted to. The inspector took the filling plug out of the fourth one, slid a long blade in, and it went the full depth of the bore without touching anything but conductor insulation and air. He then opened nine more. Six were empty. The building had been electrically complete for five months.

Nothing about the outside of a sealing fitting tells you whether it was ever filled. The body goes in with the pipe, at rough-in, usually by whoever is threading conduit that week. The compound cannot go in until the conductors are pulled, dressed and separated, which is a different visit, sometimes a different subcontractor, and always after the part of the job that anyone is watching. A seal is the one component on a hazardous-area installation that is installed twice and inspected once, and the second installation is the one that does the work.

It helps to be blunt about what that second visit is for. The compound in a seal-off is not weatherproofing, not a strain relief, not a moisture barrier and not a way of tidying an entry. It is a plug that stops a flame front travelling along the inside of a pipe, and it stops pressure travelling with it. Everything else on this page — where the fitting is permitted to sit, how many conductors may pass through it, how deep the compound has to be, what order the work happens in — follows from that single job.

What a flame front does once it is inside a pipe

An explosionproof enclosure does not keep the atmosphere out. It expects the surrounding gas or vapour to get in, expects an internal component to ignite it eventually, and is built to survive that. Its joints are machined flame paths: a defined gap over a defined length, so the burning products forced out of the box are cooled below the ignition temperature of the mixture outside before they clear the joint. That is the whole trick, and it is tested rather than calculated — UL 1203, Explosion-Proof and Dust-Ignition-Proof Electrical Equipment for Use in Hazardous (Classified) Locations, in North America, and IEC 60079-1, Equipment protection by flameproof enclosures 'd', for Ex d equipment elsewhere.

Gas grouping exists because that quenching gap is not the same for every substance. The NEC's Groups A through D and the IEC's IIA, IIB and IIC sort atmospheres by how narrow a slot will still let a flame through, which is measured as the maximum experimental safe gap. Hydrogen and acetylene sit at the demanding end; that is why a Group IIC or Group B enclosure is a heavier casting with longer flanges than a Group IIA or Group D one. It is also why nothing on this subject generalises: the group on the classification drawing is part of the specification of every fitting on the run.

Now put a threaded conduit into the side of that enclosure. It is a smooth, straight bore, orders of magnitude wider than the gap the machined joints hold to, with nothing anywhere along it to quench a flame, and it is full of the surrounding atmosphere because a conduit system breathes — it warms, it cools, it draws in whatever the room is holding. An internal ignition that the enclosure itself would have survived now has a clean route out of it, along the pipe, into the next enclosure or into a room the classification drawing says should never see a flame. The enclosure did its job and the pipe undid it.

The more expensive failure is pressure piling. When the flame front runs down a connected volume, it drives the unburnt mixture ahead of it and compresses it. The charge waiting in the second enclosure ignites from a pre-compressed starting pressure, so the pressure rise it produces is a multiple of what the casting was hydrostatically tested against. Enclosures split at bolt lines rather than at flame paths when this happens. A seal breaks the connected volume into pieces small enough that this cannot build, which is why the code cares where the fitting sits and not merely that one exists.

What is inside the chamber, bottom to top

The fitting body is the only part chosen from a catalogue. It has to be listed for the class, group and division it sits in — a fitting good for Class I Group D is not automatically good for Group B — and it goes on to threaded rigid metal conduit or threaded steel intermediate metal conduit with the five fully engaged threads that NEC 500.8(E) asks of threaded joints in explosionproof systems. Tapered thread cut short, or made up with a pipe wrench onto two threads because the run came up long, is a defect in the flame path before the compound is even mixed.

The two things poured into it do different jobs and get confused constantly. The fibre dam is packing: loose mineral fibre worked into the bore and, critically, between and around every individual conductor, so that the compound poured on top of it stops where it is meant to stop. It is not the seal. The compound is the seal, and it is a mineral powder mixed with water to a heavy cream, poured, and left to set hard against the bore wall and against each conductor's insulation. If the dam was packed around the outside of a taped bundle rather than between the conductors, the compound sets around a bundle too, and the flame path runs through the interstices between the cables. The fitting will look, weigh and sound exactly like a good one.

Where the fitting is allowed to sit

Article 501.15 of NFPA 70, the National Electrical Code, asks for seals in two quite different situations, and treating them as one rule is where most field arguments start. One protects an enclosure. The other protects a boundary. They have different distances, different lists of what may sit between the seal and the thing it is protecting, and different consequences when they are wrong.

The enclosure seal is the tight one. In a Class I Division 1 location a conduit seal goes within 450 mm — 18 inches — of any enclosure containing arcing or sparking parts, and within the same distance where trade size 2 conduit or larger enters an enclosure holding terminals, splices or taps. Between that fitting and the enclosure the code permits only explosionproof unions, couplings, reducers, elbows and capped elbows no larger than the trade size of the conduit. No condulet, no box, no length of flexible connection, and no run of pipe long enough to be convenient. A single seal is permitted to serve two enclosures joined by a nipple or a short run of conduit — not more than 900 mm, 36 inches — provided it sits within 450 mm of each of them, which is worth designing for on a pair of devices mounted close together.

The boundary seal is the loose one, and it is the one people forget entirely. Where a conduit leaves a classified location for an unclassified one, the classification does not stop at the wall — the pipe carries the atmosphere across with it. The seal may sit on either side of the boundary within 3.05 m, ten feet, and the run between the seal and the point at which the conduit crosses out must contain no union, coupling, box or fitting other than listed explosionproof reducers at the conduit entry. That last clause is the one that catches retrofits: someone adds a coupling to extend a run and turns a compliant boundary seal into a non-compliant one without touching the seal at all.

There are markings and there are exceptions, and both are worth reading rather than remembering. Equipment marked to indicate that a seal is not required at its entries has been tested that way and its nameplate is the listing speaking; a factory-sealed device does not get a second seal because someone liked the look of it. The edition of the code in force also carries exceptions for runs that pass straight through a Division 2 area and for boundaries into outdoor unclassified space, and those move between editions and local amendments. Take them off the adopted edition on this job, not off the one you learned the trade on.

Dust is a different problem wearing the same fitting name

Class II locations — combustible dust — use the word 'seal' for a job that is not the same job. Article 502.15 is about keeping dust out of the raceway and out of the enclosure at the other end of it, rather than about arresting a flame front in a bore, and the difference shows in what it will accept: alongside a compound-filled fitting it recognises lengths of raceway themselves as a seal, horizontal or vertical, and it does not require the seal to be explosionproof. A crew that installs a Class I fitting and pours it in a Class II location has done more than was asked; a crew that reasons the other way round has done a great deal less.

What does carry across is the classification discipline. The area drawing that names the class, division or zone, the group, and the temperature class comes from a study — NFPA 497 for flammable gases and vapours in chemical process areas, NFPA 499 for combustible dusts, API RP 500 and API RP 505 for petroleum facilities — and it is a document with an author and a revision, not a judgement made on the walk-round. Dust brings its own governing figure in the form of the layer and cloud ignition temperatures that set the equipment's T-code, which is a different constraint from the gas group but arrives from the same drawing. Read it before choosing a single fitting.

The same job outside the Division system

Half the world classifies by zone rather than by division and reaches the same requirement by a different route. Under IEC 60079-14, Electrical installations design, selection and erection, the conduit systems that do exist are required to carry sealing devices at the entry to flameproof enclosures and where a conduit crosses out of a hazardous area, on the same logic that Article 501.15 uses: break the connected volume, and stop the flame path where the classification changes. Article 505 of the NEC and API RP 505 provide the zone vocabulary inside North America, CSA C22.1 Section 18 does it in Canada, and AS/NZS 60079.14 in Australia and New Zealand.

The bigger practical difference is that conduit is uncommon in IEC practice at all. Cable with a certified gland is the norm, and the equivalent of a poured seal there is the barrier gland: a compound-filled or resin-filled gland, certified as an assembly, whose barrier is made up on site with a two-part compound to the gland maker's instructions. It fails the same way seal-offs do — unfilled, or filled around the outside of a bundle instead of between the cores — and it is inspected for the same reason. NEC 501.15 covers the parallel cable cases in Divisions 1 and 2 for anyone working to the American code with cable rather than pipe. Wherever the job sits, the duty behind it is statutory: DSEAR in Great Britain, with its Approved Code of Practice, and the ATEX workplace directive across the EU.

Proving, later, that a seal was ever poured

Nothing in Article 501.15 tells you to mark a poured seal. That is trade discipline rather than code, and it is worth adopting anyway, because the alternative is the afternoon this page opened with. Number every fitting on the drawing, tick it off in a seal schedule when the compound goes in, sign the line, and photograph the open chamber with the compound visible before the plug goes back. A painted plug or a stamped tag gives an inspector something to sample against rather than a reason to open all forty-one.

That record is also the thing that survives the crew. Hazardous-area installations are subject to a periodic inspection regime — IEC 60079-17, Explosive atmospheres, electrical installations inspection and maintenance, sets out initial, periodic and sample inspection in the IEC world, and the equivalent duty sits with the operator everywhere else. An inspector arriving in year four with a seal schedule can verify a sample and move on. An inspector arriving with no schedule has to treat every fitting as unknown, and the cost of that lands on the operator rather than on whoever saved an hour not writing it down.

The last thing to hand over is the constraint itself, in writing. Every future modification to a sealed run reopens the question this page answers: a conductor added, a device swapped for one whose nameplate no longer says the seal is not required, a coupling installed on the wrong side of a boundary seal, a fitting relocated more than 450 mm from the enclosure it protects. Those are the changes a maintenance team makes in an afternoon without knowing what they were touching. A drawing that marks the classified boundary and the seals on it is what stops that, and it is cheaper than the alternative by several orders of magnitude.