A hotel project in Doha taught me to look twice at the outside of a double-wall duct. The ducts were fine on paper. Insulation thickness right, perforation pattern right, leak test passed. Then the installation crew started handling the sections, and about 15 percent of the outer skins arrived on the deck with dents you could see from ten metres away. The client called it poor manufacturing. It was a gauge decision made on price.
Most of the technical conversation about double-wall duct is about the liner: perforation percentage, acoustic blanket, cleaning and inspection. That side matters, and we covered it elsewhere. The outer jacket gets far less attention, and it is the part that takes the beating, carries the flange, and decides whether the duct still looks straight after three months on site.
In a double-wall duct the inner skin is perforated and the space between the skins is filled with insulation. The inner layer manages airflow and noise. The outer layer is the structural skin: it keeps the section square, carries the flange at each end, provides the vapour barrier for the insulation, and survives transport and handling.
That definition changes how you pick the material. A single-wall duct skin can be chosen on pressure and span alone. A double-wall jacket has to be stiff enough to stop the section from oil-canning when a man stands on it during installation, and smooth enough that a dent does not telegraph through the insulation to the liner.
Pressure drop across the wall is small, so the static pressure class barely affects the jacket gauge. Width does. A 300 mm wide panel at 0.6 mm feels rigid because the panel is narrow. The same 0.6 mm sheet at 1,250 mm wide will drum like a tin can the moment someone leans on it.
| Duct width | Jacket gauge, galvanised | Beading pattern | Max hanger spacing | Practical note |
|---|---|---|---|---|
| 200–400 mm | 0.6 mm | 1 bead | 2.4 m | Handles two men standing on it |
| 450–650 mm | 0.8 mm | 2 beads | 2.4 m | Most common specification |
| 700–1,000 mm | 1.0 mm | 2–3 beads | 2.0 m | Needs a rigid flange or it twists |
| 1,050–1,250 mm | 1.2 mm | 3 beads | 1.8 m | Two-man Handling, corner brackets required |
| Above 1,250 mm | 1.2 mm + angle frame | 3–4 beads | 1.5 m | Beading alone is not enough |
The beading is doing structural work here. A 0.8 mm jacket with two 8 mm beads across a 600 mm face is appreciably stiffer than a flat 1.0 mm jacket, and it uses less steel. On our line the beads go in before the flanges are formed, because a duct beader cannot roll a bead through a finished flange profile without collapsing the pocket.
Moving a 600 mm jacket from 0.8 to 1.0 mm adds about 25 percent to the sheet weight of the jacket, roughly 2.1 kg per metre of duct. On a 3,000 m job that is 6.3 tonnes of extra steel. Against that, the damage rate on site might fall from 15 percent to 3 percent, and each damaged section costs a day of rework plus a crane slot to replace it. The heavier gauge is nearly always cheaper by the time the duct is installed.
I now quote both options and mark the lighter one as "shop-handling only". That phrase has saved me a lot of arguments.
Galvanised steel is the default, but not every coating is the same and not every site suits it.
For a normal interior HVAC system, Z275 galvanised at 0.8 mm is the workhorse. The coating gives you 275 g of zinc per square metre across both faces and stands up to years of handling and the occasional wet service. The price premium over Z140 is small enough that I rarely bother quoting the lighter coating for jackets.
In airports, hospitals and food plants the jacket is often exposed, and the client wants 304 stainless. It is a different fabrication job: the material work-hardens, the rolls mark it more easily, and the bead depth has to be shallower to avoid cracking. Expect 8 to 12 percent scrap on the first batch until the operators settle, and protect the surface with film from the coil to the wall. A single bare steel tool mark on a stainless jacket will rust in a humid plant room and the client will see it every day.
Aluminium jacket at 1.0 mm weighs about a third of galvanised steel, which matters on a ship or a retrofit where the structure was not designed for the duct load. It also dents more easily and it galvanically corrodes where it touches a steel flange or a steel hanger. If you use aluminium, use aluminium anchors, and isolate every steel fixing with a washer.
Painted colour-coat sheet is a specialist choice. It looks good in an open-plan ceiling and it is unforgiving: one scratch through the coating and the steel underneath starts to rust. I only use it when the architect insists and the access is restricted.
The outer skin has to be continuous enough to keep moisture out of the wool. That means the seams and the flange interface need attention, and it means the fixings should not create a cold bridge from the duct surface to the room.
The jacket closes with a locked seam on the long edge, the same Pittsburgh profile used on single-wall duct, and the seam gets a continuous bead of sealant on the inside before closing. Applying it after the lock is closed is a waste of sealant. The lock former needs its rolls set slightly wider for coated jacket sheet, because the coating adds thickness without adding stiffness and a tight roll will imprint the finish.
At the corners, use mechanical corner brackets with a gasket rather than welding the jacket. Welding burns the zinc, leaves a discoloured patch, and conducts heat straight through the insulation layer. A stamped bracket with a 3 mm closed-cell gasket costs less and seals better.
Inside the cavity, the insulation is held by welded or glued pins. On a 600 mm wide panel, pins at 300 mm centres in both directions are enough. Fewer than that and the blanket sags after a year, leaving a void at the top of the cavity that shows up as a cold spot and, on a chilled system, condensation inside the jacket. If the duct is over 1,000 mm wide, add a centre row.
Check the pins before closing the jacket. It is the last time anyone will see them, and a pin that was never welded is impossible to fix from outside.
Double-wall sections get dented in three places: on the line at the outfeed, in the stack, and on the truck. The outfeed is easiest to fix. A roller table at the right height and a soft stop prevent the leading edge from hitting a steel stop every cycle.
In the stack, never put a heavy section on a light one. Sort by width, stack with 25 mm timber spacers, and keep no more than six sections high. I have seen a stack of eight 1,250 mm sections where the bottom two were bowed by the weight of the top six, and no amount of straightening brought them back.
On the truck, wrap the top and sides with a breathable cover. Shrink wrap looks professional and traps moisture against the galvanised surface; on a sea voyage it produces white rust across the whole jacket. If the duct goes by container, add desiccant and leave an air gap around the stack.
Flatness. A straightedge across the face at 45 degrees. Gap over 3 mm at 600 mm width means the beading is wrong or the section was stacked badly.
Flange squareness. 2 mm maximum across the face. A twisted flange on a double-wall section is worse than on single-wall, because the insulation cavity does not let you pull the corner back into shape on site.
Coating continuity. Look at the bead rolls and the lock seam for scratches. A scratch that reaches bare steel gets a touch-up coat of zinc-rich paint before despatch.
Liner alignment. The perforated skin should sit flush with the flange pocket so the joint is smooth on the air side. A liner 5 mm proud catches the gasket and creates a leak path at every joint.
On a chilled-water system the jacket is also the vapour barrier. If warm room air reaches the cold insulation through a seam or a pin hole, moisture condenses inside the cavity and stays there. Six months later the bottom of the duct is wet, the insulation has lost its value, and the outer skin is rusting from the inside. The failure looks like a coating problem but it is almost always a sealing problem.
That changes the standard for the jacket seams. Every longitudinal lock gets sealant inside before closing. Every flange corner gets a gasket, not just a dab of mastic on the outside. And every insulation pin that penetrates the jacket outer face gets a sealed washer. On a run that carries 7 °C air through a 28 °C plant room, there is no such thing as an unimportant pinhole.
A 50 mm blanket on a 600 mm duct gives the jacket some support; a 25 mm blanket on the same duct lets the panel flex more under a hand. If a client asks to cut insulation thickness to save money, expect to add a bead or move up a gauge to keep the panel feeling solid. The two decisions are not independent, and quoting them separately is how shops end up with a duct that passes every test and still gets rejected on site.
Where the duct runs outdoors, add a 0.2 mm polythene vapour layer outside the insulation and under the jacket, with laps of at least 100 mm sealed with compatible tape. It costs very little and it is the difference between a jacket that lasts fifteen years and one that needs replacing in five.
Yes, with care. The jacket is usually a heavier gauge than the liner, so set the roll gaps for the thicker sheet and run a sample through both skins before the batch. Expect one extra adjustment per changeover.
Usually the bead pattern is too sparse for the width, or the bead depth is under 6 mm. On a 1,000 mm face, three beads at roughly 250 mm spacing with 8 mm depth is the minimum I would accept.
Yes. A generic acetic-cure silicone will corrode aluminium and stain stainless. Use a neutral-cure silicone or a butyl mastic that is rated for both metals, and check the gasket is compatible too.
Not if anyone has to handle the section. It will pass a pressure test in the shop and arrive dented. Save 0.6 mm for narrow branches or for duct that is assembled and insulated in place, where the jacket has continuous support.
If you are setting up the line for this kind of work, the throughput figures for the five-station configuration are a useful starting point; double-wall sections take roughly 20 percent longer per metre because of the extra handling, not because the forming is slower.