The myth that roll forming is only bending
I'm Iris Xu, a Sales Manager in ZTRFM's Business Department 6. A common myth on the factory floor is that roll forming is just bending a flat strip into shape. It is not. Alongside the transverse bend, the strip also stretches lengthwise at its edges and thins across its web. Ignore those two effects and you get edge wave, oil-canning, or split corners. They are the reason a profile that looks simple on a drawing can still fail on the line.
Start with the edge. When you form a tall rib or a deep flange, the material at the very edge has to travel a longer path than the material at the centre of the web. Picture a flat blank of half width W entering as a straight line and leaving as a section where the edge is now raised. The line along the edge is simply longer than the line down the middle. To fill that longer path the edge must elongate. That elongation is the longitudinal edge strain, and it is what forms the corner radius in the first place.
The amount of strain is not random. Research and our own flower software use a useful approximation: the peak longitudinal strain at the edge scales with the square of the flange width and the square of the per-stand bend increment, and drops with the square of the inter-stand distance. In plain terms, wide flanges strain far more than narrow ones, and spreading the same bend over a longer machine reduces the peak. That is why production lines run inter-stand distances of 250 to 600 mm and why each stand only turns the flange a few degrees.
| Lever | Effect on peak edge strain |
|---|---|
| Wider flange (b) | Strain grows with b squared |
| Bigger bend step (delta phi) | Strain grows with step squared |
| Longer inter-stand gap (L) | Strain drops with L squared |
When the edge gives up
If the edge strain stays below the elastic limit, the edge springs back and the panel is straight. Push it past yield and the edge stays permanently longer than the centre. The excess length has nowhere to go, so it buckles sideways into an edge wave. High-strength steels are worse here because they have a lower n-value, the work-hardening exponent that lets a metal spread strain evenly. A low n-value means strain concentrates at the edge instead of spreading, and the wave appears sooner.
The free edge is the weak point. It is not held by material on either side, so under compression it buckles out of plane; under tension it can thin and even tear. Good pass design gives the edge lateral support through side rollers and guides, so it cannot wander while it is being formed.
Web thinning
The web is the flat middle of the section, and it thins under pressure too. In the bend region the material is squeezed between top and bottom rolls, so the wall there is slightly thinner than the flat. Away from the bends the strip keeps its original thickness. Thinning is usually small, a few percent, but it matters for coated steel. A galvanized coating is a fixed mass of zinc per square metre, so where the base steel thins the coating proportion goes up, and a tight bend can even crack the coating if the radius is too small.
Web thinning also shows up as oil-canning. If the centre of the web yields under the reactive compression from the flanges climbing, it buckles into a poppable, oil-can shape. Like edge wave, it is a longitudinal-strain failure, not a transverse one. Wider flats need more stands and gentler steps for the same reason wide flanges do.
How the flower controls it
The flower design is the map that decides how much bend happens at each stand. A good flower keeps the peak longitudinal strain under the material's limit at every step, balances the edge against the centre, and brings the section to full shape only at the last stand. A poor flower piles too much angle into the early stands, spikes the edge strain, and the first panel out of the line is wavy.
In practice we check three things before releasing a new profile:
- Peak edge strain versus the grade's allowable longitudinal strain.
- Coating thickness at the tightest bend, to avoid zinc cracking.
- Web flatness margin, so oil-canning does not appear after springback.
A production example worth remembering
Last year a customer in Vietnam sent us a 1.6 mm high-strength section, S550GD, with a 40 mm flange and a 6 mm lip, running at a 250 mm inter-stand distance. Our software flagged the peak edge strain at about 4.5 percent, past the 3.5 percent that S550GD tolerates before edge wave. We widened the inter-stand gap to 450 mm and added two stands, which dropped the peak to 2.8 percent. The first trial run after that change came out straight. No new rolls, no new steel, just a longer, gentler flower. That is the whole point of controlling longitudinal strain instead of fighting it after the fact.
For coated strip the coating math matters too. A Z275 coating carries 275 g of zinc per square metre. When the base steel thins 5 percent under the bend, that same mass now covers less steel, so the coating effective thickness rises about 5 percent locally. Usually fine. But at a bend radius tighter than three times the thickness, the zinc can fracture and the steel underneath shows. We therefore cap the tightest bend on galvanized stock at an R/t of 3 unless the customer accepts a touched-up corner.
We also measure thinning directly. After a run we cut a cross-section, polish it, and read the wall at the bend against the flat with a micrometer. A healthy profile shows under 4 percent thinning at the bend and nothing across the web. If the web reads thinner than the original coil, the reactive compression was too high and we add a flattening stand near the exit. It is a five-minute check that prevents a complaint three months later.
What we tell buyers who see waves or oil-canning:
- Send the grade and the n-value, not just the thickness. Low n is the usual suspect.
- Widen the inter-stand gap before you slow the line. Spreading the bend is cheaper than re-rolling.
- Check the coating at the tight bend. Zinc cracks show as bright spots after forming.
- Let the flower, not the operator, set the per-stand angle. Consistency beats feel.
Takeaway
Roll forming is bending plus stretching plus thinning, and the stretching and thinning are where profiles actually fail. Size the flange, the inter-stand gap, and the bend step together, not separately, and the edge and web will stay inside their limits. A wave-free panel is a flower-design problem long before it is a machine problem.
At ZTRFM we run the flower calculation for every profile we ship, so the edge strain and web thinning are planned, not discovered on the floor. If your line is throwing edge waves on a high-strength section, send us the strip width, flange height, and steel grade and we will check whether the pass plan is the cause.
