Most of the duct lines I commission in Southeast Asia and the Middle East cut rectangular blanks with a shear. It is fast, it leaves a clean edge, and it costs almost nothing to run. So when a customer asks whether he should pay extra for the CNC plasma station, my first answer is usually no. Then he tells me he also builds transitions, offsets, and odd-shaped fittings, and the answer flips.
A shear cuts straight lines only. Every time your shop needs a taper, a notch that is not square, a radius, or a hole for a damper shaft, someone walks the part over to a hand plasma torch or a band saw and the part sits there for twenty minutes. On a line running 800 to 1,200 m² of duct a month, that walking time is usually where the profit leaks out, not in the cutting itself.
What follows is what I have learned setting up plasma stations on 1.25 m and 1.5 m wide lines: the numbers that work on galvanized coil from 0.6 to 2.0 mm, the three kinds of dross you will see and what each one is telling you, and how long consumables actually last when you are cutting coated steel all day.
People spend weeks comparing torch brands and five minutes deciding where the table sits. That is backwards. The torch is a commodity. The integration is what determines whether the station runs or fights you.
On a full auto line the order is normally decoiler, straightener, bevel and notch station, then either a shear or a plasma table, then the lock former, then the folding station. Putting plasma after the shear sounds redundant, but it is the configuration I recommend for shops that build a lot of fittings. The shear handles the 80 percent of parts that are plain rectangles at full speed, and the plasma handles the rest without a changeover.
If the plasma is the only cutting station, put it directly after the straightener and give it at least 2.5 m of outfeed table before the lock former. Blanks that drop onto a short table get dragged, and a dragged blank has a burr on the leading edge that shows up later as a scratch on the folding blade.
Cutting galvanized sheet produces zinc oxide fume, and a little of it goes a long way. I insist on a downdraft table or a water bed under the cutting zone. A water bed is cheaper and quieter, but it makes the slugs wet and they stick to the blank, so the operator ends up wiping parts. A downdraft table with a 2,000 m³/h fan and a pleated filter is the setup I prefer. Budget for filter replacement every four to six months in a two-shift shop.
Leave 1.2 m clear behind the table for the extraction duct and 800 mm on the operator side. Do not put the electrical cabinet within 1.5 m of the cutting zone. I have replaced two drive boards in five years because conductive dust found its way into a cabinet that was mounted too close.
The table below is what I start with on a 63 A air plasma torch with a 1.1 mm nozzle, cutting hot-dip galvanized coil with a zinc coating around 120 to 275 g/m². Treat it as a starting point, not gospel. Every torch has its own sweet spot, and the only way to find it is to cut a test strip and look at the bottom edge.
| Sheet thickness | Current | Cut speed | Cut height | Air pressure | Kerf | Typical use |
|---|---|---|---|---|---|---|
| 0.6 mm | 30 A | 4,500 mm/min | 1.0 mm | 0.45 MPa | 1.2 mm | Small fittings, damper blades |
| 0.8 mm | 40 A | 3,800 mm/min | 1.2 mm | 0.50 MPa | 1.4 mm | Standard low-pressure duct |
| 1.0 mm | 45 A | 3,200 mm/min | 1.5 mm | 0.55 MPa | 1.5 mm | Most rectangular ductwork |
| 1.2 mm | 60 A | 2,600 mm/min | 1.5 mm | 0.60 MPa | 1.7 mm | Medium-pressure, 800 Pa class |
| 1.5 mm | 80 A | 1,900 mm/min | 1.8 mm | 0.60 MPa | 1.9 mm | Heavy gauge, industrial exhaust |
| 2.0 mm | 100 A | 1,300 mm/min | 2.0 mm | 0.65 MPa | 2.2 mm | Supports, hangers, thick transitions |
Two numbers in that table cause more trouble than the rest put together: pierce height and cut height. Pierce height should be two to two and a half times the cut height, so 3.0 to 3.5 mm on 1.2 mm sheet. Pierce too low and the molten metal blows back into the nozzle, which is how you burn a nozzle in an afternoon. Pierce too high and you spray spatter across the top surface, and spatter on galvanized sheet is almost impossible to remove without damaging the coating.
Air quality is the other silent killer. Air plasma wants clean, dry air at a dew point below minus 20 degrees Celsius. A shop running a piston compressor without a refrigerated dryer will go through electrodes three times faster than the book says, and the operator will blame the torch. Put a coalescing filter right at the machine inlet and drain it every shift. It costs less than two nozzles.
Dross is not a single problem, and the fix depends entirely on which one you have. I teach operators to look at three things: where the dross sits, what it looks like, and whether it comes off with a fingernail or a grinder.
Heavy, bubbly, blobby dross on the bottom edge, hard to chip off. This means the torch is moving too slowly or the arc is too hot for the thickness. The molten kerf never gets blown clear and it solidifies on the underside. On 1.0 mm galvanized at 45 A, dropping from 3,200 to 2,700 mm/min will produce this within a metre. Speed up in 10 percent steps until the dross turns into a thin, powdery line, then back off 5 percent.
Thin, fine, tightly attached dross that leans in the direction of travel, often with a rounded top corner. This is the opposite problem: the arc is running ahead of the cut and not penetrating fully. Slow down, or check that the nozzle orifice is not worn oval. A nozzle that has cut more than about 900 pierces on 1.2 mm sheet will start producing this no matter what you do with speed.
Beads of metal sitting on the top surface, usually clustered around pierce points and sharp corners. Ninety percent of the time this is pierce height, not cut height. The other ten percent is a dirty shield cap or a missing shield gas swirl ring. If the spatter only appears at corners, look at the lead-in. A 3 mm lead-in on a 90 degree corner is not enough. Use 6 to 8 mm, and start the lead-in on the scrap side so the pierce mark never lands on the finished edge.
Buyers ask me whether a plasma edge is good enough for a Pittsburgh lock. It is, if you keep the cut dry and the nozzle fresh. A plasma cut on 0.8 mm galvanized typically shows a bevel of 3 to 5 degrees and a heat-affected zone of 0.1 to 0.25 mm. The edge is square enough for lock forming, but it is harder than a sheared edge, and that matters at the lock former.
Hard edges wear the forming rolls faster and they crack more readily at tight bend radii. On a Pittsburgh lock with a 6 mm hammer-over, a plasma-cut edge will occasionally show a hairline crack at the corner of the female leg after forming. Two fixes work. First, run the plasma slightly on the hot side so the HAZ is shallow rather than deep, which sounds counterintuitive but gives a finer grain structure. Second, keep the lock former rolls clean and check the clearance; a roll with 0.15 mm of play will crack a hard edge that a sheared edge would survive.
If the part goes straight to folding with no lock, plasma is fine and often better, because there is no shear burr to mark the folding blade. For visible architectural duct where the raw edge will be seen, neither process is acceptable and you should be planning a folded hem anyway.
On air plasma cutting 0.8 to 1.2 mm galvanized in a two-shift shop, I plan on the following: electrode and nozzle every 900 to 1,200 pierces or roughly 3 to 4 hours of arc-on time, shield cap every three nozzle changes, swirl ring every six months. That works out to an electrode and nozzle set every one to two weeks per shift.
Where shops get this wrong is buying the cheapest consumables they can find. A set of genuine electrodes costs maybe four times a no-name set, and lasts three to four times as long while holding a consistent kerf. The hidden cost of the cheap set is not the price. It is the operator stopping to re-cut parts, and the fitter downstream discovering that the 1.4 mm kerf he designed around is now 1.9 mm.
Keep a log taped to the machine: date, thickness, pierces, and a note on edge quality at the end of the run. After two months you will know your real consumable cost per square metre, and you can quote fittings with actual numbers instead of guessing.
A plasma table without nesting discipline wastes more material than a shear ever will, because the software will happily place parts wherever they fit. Three rules cover most of it.
On a 1.25 m coil, the practical difference between a well-nested fitting program and a hand-laid one is usually 8 to 12 percent of material. On 20 tonnes of coil a month, that is the difference between a good quarter and a bad one. If you want to see how the cutting station is normally integrated on a full line, the Line 5 auto duct line page shows the standard arrangement, and the CNC plasma cutting machine page covers the standalone version for shops that already have a shear.
If you are ordering a line this year and plasma is on the option list, here is my honest shortlist of what is worth paying for. A 1.5 m wide table rather than 1.25 m, because coil availability in 1.5 m is better in most export markets and the extra width costs less than a second machine later. A downdraft table with a real filter, not a fan blowing into the shop. Torch height control that senses the plate, because on 0.6 mm sheet a manual head will not hold tolerance. And a nesting software licence that the operator can actually use, which in my experience means one with a Chinese and an English interface, because the person running the night shift is rarely the person who bought the software.
Skip the high-definition torch unless you are cutting stainless above 3 mm. On galvanized duct sheet it buys you nothing you can measure, and it doubles your consumable bill.
Yes, but not with the same consumables or the same gas. Stainless needs either nitrogen or a nitrogen-hydrogen mix to avoid an oxidized edge that will not weld cleanly, and the cut speed drops by roughly 25 percent at the same amperage. Keep a separate set of consumables marked for stainless, and purge the line before you switch back.
Within about 1 mm over the flange height, so roughly 1.5 degrees of bevel on a 40 mm flange. Above that, the flange roll will pull the blank off square and you will see it as a gap at the corner insert. If your bevel is consistently over 2 degrees, the nozzle is worn or the cut height is too high.
Usually the blank is moving, not the torch. Check the clamping on the outfeed side, and check for a slug that has dropped onto the table and lifted the sheet. On thinner material, heat distortion will also push a long narrow part sideways; cut it in two passes from the middle out rather than one continuous line.
Water captures fume better and costs less to install, but wet slugs stick to blanks and the water turns into a zinc-rich sludge you have to dispose of. In a shop with no local exhaust regulation, water is the pragmatic choice. Where there is any enforcement, go downdraft.