Every shop I visit in the Gulf or West Africa has a plastic tub of TDF corners somewhere near the flange machine. Usually it is half empty, and someone is on the phone asking for air freight on a part that costs eight cents.
That tub is the whole argument. Corners are small, cheap, boring, and absolutely critical: one missing corner stops a duct from being assembled, and a corner made to the wrong tolerance cracks the flange it is pressed into. So the question of making them yourself is not really about the eight cents. It is about whether you can afford to be four weeks away from a stockout.
I have set up corner production in three shops and advised against it in four others. Here is the arithmetic I use.
A TDF corner insert is a folded steel bracket that joins two flange profiles at a 90 degree corner and locks them with a bolt. The flange itself is formed by the rolls on the TDF flange forming machine, which turns the duct edge into a 20 or 30 mm pocket. The corner slides into two pockets and a bolt pulls the joint tight.
The part looks trivial and is not. The critical dimensions are the leg length that engages the flange pocket, the height that matches the flange profile, the hole position, and the flatness of the face that seats against the gasket.
Standard corners are stamped from 1.2 to 1.5 mm galvanised strip, 62 to 75 grams each depending on the flange height. Heavy-duty versions for large duct run to 2.0 mm and 110 grams. The tolerance that bites is leg length: plus or minus 0.3 mm is normal, and anything beyond that either rattles in the pocket or will not enter it.
Hole diameter is usually 8.5 mm for an M8 bolt, drilled or punched with a plus tolerance of 0.2 mm. Hole position matters more than diameter, because two corners on the same joint have to line up with the bolt passing through both.
Then there is flange height. A 30 mm corner in a 20 mm flange will bottom out before the joint closes and crack the flange roll at the corner. Shops that stock one size and use it everywhere are the ones with cracked corners on site.
The ex-works price for a standard 20 mm galvanised corner from a Chinese or Turkish supplier runs USD 0.07 to 0.09 at 50,000 pieces, and USD 0.06 at container volume. That is not the number that matters.
Landed cost adds freight, duty and the cost of holding stock. A 20 foot container holds roughly 2.6 million corners in cartons, about 1.3 tonnes. Sea freight on that from Shanghai to Lagos, at rates I have seen in the last two years, adds USD 0.010 to 0.018 per piece. Duty in most West African markets is 10 to 20 percent, plus clearing and inland haulage. Realistically you land a standard corner at USD 0.10 to 0.11, and you land it six to twelve weeks after you order it.
That lead time is the expensive part. A shop running 20,000 corners a month cannot hold more than about eight weeks of stock without dedicating a rack to it, and eight weeks of stock is USD 17,000 sitting still. When a container is delayed at Apapa for a month, the line stops or someone pays air freight at USD 4 a kilogram for a part that weighs 62 grams.
Making corners is a stamping job, and stamping economics are driven by three things: material, strokes per hour, and die life.
A standard corner uses 62 grams of 1.5 mm galvanised strip. At a coil price of USD 800 a tonne, that is USD 0.050 of material per piece before scrap. With 8 percent scrap on a progressive die, the material cost per good piece is about USD 0.054. That number barely moves with volume, which is why corner making never becomes dramatically cheaper than buying material.
A single-station punching machine with a manual feed will do 800 to 1,000 pieces an hour with one operator, which is 20 seconds of work per 18 pieces. A dedicated TDF corner making machine with an automatic feeder runs 1,800 to 2,200 an hour. The fully automatic corner production line, which uncoils, straightens, punches, forms, and counts, runs 3,000 to 3,600 an hour with one man watching it.
At a loaded labour rate of USD 5.50 an hour, the labour content is USD 0.006 at 900 per hour and USD 0.002 at 3,000 per hour. Small numbers, but they are the whole difference between the two machine routes.
A corner die set costs USD 1,500 to 2,500 and lasts 100,000 to 150,000 strokes before sharpening, with three or four sharpenings before it is scrap. Call it 400,000 pieces over its life, which puts die cost at USD 0.005 per piece. Sharpening takes two hours and a die grinder, and if you cannot do it in-house you send it out for a week.
Punch-to-die clearance should be about 8 percent of thickness, so 0.12 mm on 1.5 mm strip. Too tight and the die wears at the corners; too loose and you get a burr that cuts the operator's glove and scores the flange pocket.
Below is the monthly cost at four volumes. Bought figures are landed cost at USD 0.105 per piece. Made figures include material, labour, die amortisation, power and 2 percent internal reject.
| Corners per month | Bought, landed | Made on punch press | Made on auto line | Annual saving, press | Annual saving, auto |
|---|---|---|---|---|---|
| 5,000 | USD 525 | USD 380 | USD 345 | USD 1,740 | USD 2,160 |
| 20,000 | USD 2,100 | USD 1,520 | USD 1,380 | USD 6,960 | USD 8,640 |
| 50,000 | USD 5,250 | USD 3,800 | USD 3,450 | USD 17,400 | USD 21,600 |
| 100,000 | USD 10,500 | USD 7,600 | USD 6,900 | USD 34,800 | USD 43,200 |
Two things jump out. The saving per piece is roughly 27 to 35 percent, which is real but not transformative. And the absolute saving at 5,000 a month is USD 145, which does not justify a machine, a die, an operator and a maintenance burden.
A single punching machine with a feeder is USD 6,000 to 9,000 installed. At 20,000 corners a month the saving is USD 580, so payback is 13 to 15 months. At 50,000 a month it is five to six months, and that is a clear yes.
An automatic corner line is USD 28,000 to 42,000 depending on whether it includes the uncoiler and the counter. At 50,000 a month the saving over buying is USD 1,800, so payback is 16 to 23 months. At 100,000 a month it is eight to eleven months.
My rule: below 30,000 corners a month, buy. Between 30,000 and 60,000, a punch press with a feeder is defensible if you also value the lead time. Above 60,000, an automatic line pays for itself inside two years and the lead-time benefit is free.
The payback maths assumes the machine runs. Three things have to be true.
Corner strip is usually 1.5 mm galvanised at 60 to 100 mm width. If you are already buying coil for the duct line, slitting a 1,250 mm coil down to corner strip on your own slitter is the cheapest route, and it removes a supplier. If you are buying pre-slit strip in small quantities, you will pay a premium that eats a third of the saving.
Stamping dies fail in predictable ways: worn punches, chipped corners, stripped stripper plates. A shop that cannot sharpen a punch will end up with a burr problem within 40,000 strokes. Budget for a surface grinder and 20 hours of training, or accept that you will ship dies out twice a year.
Twenty and thirty millimetre covers almost every job. Once a shop starts making five heights to match every customer's preference, the die changes eat the labour saving. Two sizes, two dies, one setup each per month.
Corners are the part everyone assumes is fine. Then a flange cracks on site and the argument starts. Four checks, done hourly during a run.
Leg length. Vernier across the two legs. Plus or minus 0.3 mm. A corner 0.5 mm over will not enter a tight pocket.
Hole position. Check one in fifty against a gauge plate with two pins. If the holes on a pair do not line up, the bolt goes in at an angle and the joint never pulls square.
Burr. Drag a thumbnail along the sheared edge. Anything that catches is too much. Burrs score the inside of the flange pocket, and a scored pocket is where the zinc fails first.
Fit test. Press a corner into an offcut of flanged duct by hand. It should go in with thumb pressure and stay there. If it needs a hammer, the height is wrong. If it falls out, the leg is undersize.
On galvanised corners, ask for the coating weight. A 1.5 mm strip at Z120 will show white rust in a humid store within a month if the corners are packed wet. I have opened cartons in Lagos where the corners had fused into a block. Specify dry packing with a desiccant and keep the tub off the floor.
Four situations, and they come up more often than the machine salesmen admit.
Shops under 20,000 corners a month. The saving does not cover the attention the machine needs.
Shops doing stainless and aluminium work. Stainless galls on a die set cut for galvanised, and you need different clearance and lubrication. Running both on one die is how you get torn edges and a scored die face.
Shops with no maintenance capability and no plan to build one. A press with a dull die produces 100,000 bad corners before anyone notices, and they all get used.
Shops that are cash constrained and growing. USD 30,000 spent on a corner line is USD 30,000 not spent on a second coil or a delivery van. If the line is the bottleneck, buy the line. Corners are not usually the bottleneck.
About 16 at 62 grams each, allowing for scrap. A 1.3 tonne container load of bought corners is roughly 21,000 pieces of material, which is why freight per piece looks small and stock value looks large.
Yes, with two die sets. Changeover is 20 to 25 minutes on a single-station machine and about 40 minutes on an automatic line, mostly re-threading the strip and re-setting the counter.
Not necessarily, but you need a heavier press. A 2.0 mm corner needs roughly 40 percent more tonnage than 1.5 mm on the same profile, and a light press will flex and produce a corner that is thick at one end.
Almost never. The handling cost of painting 50,000 small parts exceeds any saving, and painted corners chip in the pocket during assembly. Galvanised, dry, and correct height is the right answer.
If you are working out whether the corner route fits your shop, the useful starting point is your flange consumption: count the corners in a month, multiply by your landed price, and see whether the number is over USD 3,000. Below that, keep the tub and keep buying.