Tube bending defects are the geometric and surface failures that appear when a tube is formed around a die. Thinning, ovality, wrinkling and angular error are the four that show up on inspection reports. They are not random. Each of these tube bending defects has a dominant cause, a measurable acceptance criterion, and a fix that lives in the tooling set or the machine specification and not in operator technique.
This article is written for the person who signs off the part or specifies the machine that makes it. For each defect it gives identification, cause and equipment-side fix. It does not cover improvised workarounds or field repairs. If a bend is out of tolerance because the radius was never achievable in that wall thickness, that is a specification problem, and pushing harder on the machine makes it worse.
The Four Tube Bending Defects in One Matrix
Read this against the last batch you rejected and the defect normally names itself. Sorting rejected parts by where on the bend the fault sits settles most arguments about which of the tube bending defects you are looking at.
| Defect | Where it appears | What you measure | Dominant cause | Equipment-side fix |
|---|---|---|---|---|
| Wall thinning | Outside of the bend, worst 45 to 90° through the arc | Remaining wall by ultrasonic gauge, as a percentage of original | Outer fiber in tension, so it travels further and stretches | Open the CLR, use a following pressure die, correct mandrel support |
| Ovality | Across the bent section, peaks near mid-arc | (Dmax − Dmin) ÷ Dnominal × 100% | Unsupported section collapsing toward the neutral axis | Correct mandrel ball set, tighter die cavity fit, pressure die load |
| Wrinkling | Inside of the bend, ripples running across the tube | Ripple height against a straight edge, or a go/no-go sample | Inner wall in compression buckling instead of thickening | Wiper die fit, mandrel nose position, following pressure die, lubrication |
| Springback | Angle and plane of the part, not the bend surface | Included angle after release versus programmed angle | Elastic recovery on release, a material property and not a fault | Over-bend compensation, headroom past the target angle, consistent material |
Springback is the only one of the four that is not damage. The tube is unharmed and is simply not where you wanted it, which makes it the one defect that no amount of tooling spend will remove. And three of the four fixes point at the same hardware, the mandrel, the pressure die and the wiper die. Rotary draw bending supports the tube from several sides while it is drawn around the die, and defects appear where that support is missing. The selection logic for the hardware is covered in tube bending tooling and how to match a die set to the job, so this article stays on the diagnostic side.
Two Numbers That Decide Every Tube Bending Defect
Two ratios govern the difficulty of a bend, and between them they predict most rejections caused by tube bending defects.
CLR/OD tells you how tight the bend is. It is the centerline radius divided by the outside diameter, written as a multiple of diameter, meaning a 3D bend has a radius three times the tube OD. UK and Australian drawings call it the centreline radius. It controls how far the outer fiber must stretch and therefore how much the wall thins. Thinning is modest above about 3D and rises quickly below 2D. The radius itself is defined in centerline radius explained for tube bending.
D/t is the wall factor. It is the outside diameter divided by the wall thickness. This one controls buckling, meaning whether the section holds its shape or collapses. A Φ50 × 5 mm tube has a wall factor of 10 and is forgiving. A Φ50 × 1 mm tube has a wall factor of 50 and will flatten and wrinkle at radii the heavy wall tube shrugs off.
Counter-intuitively, a thicker wall is harder to bend in force terms and easier in quality terms. Heavy wall benders are built massive and not clever, while thin wall work needs finely fitted tooling more than tonnage.
The outer fiber sits half an outside diameter further from the bend center than the neutral axis, and so travels further, and to a first approximation the wall thins in that same proportion. The estimate is thinning (%) = 100 ÷ (2 × CLR/OD + 1).
| CLR/OD | First-order thinning estimate | Tooling implication |
|---|---|---|
| 1.5D | 25% | Mandrel and wiper die required, the working limit of rotary draw |
| 2D | 20% | Mandrel needed above a wall factor of about 10 |
| 2.5D | 16.7% | Mandrel still needed on thin wall |
| 3D | 14.3% | Mandrel optional on heavy wall |
| 4D | 11.1% | Pressure and clamp die often sufficient |
| 5D | 9.1% | Basic tooling set |
Treat those figures as an upper bound and not a prediction. A correctly set following pressure die feeds material in from behind, so measured thinning on a well-set cell runs below the table value. A machine without that assist, or with it fighting a badly positioned mandrel, meets or exceeds the value. That gap is the value of the equipment, and only a sample bend reveals it.
The rule that falls out of both ratios is the one most fabrication drawings should have started from. The minimum bend radius should not normally be less than 2 to 2.5 times the tube outside diameter. Aircraft and engine ducting goes tighter because space forces it, and pays for it in tooling and scrap. The standards side of that choice is covered in tube bend radius standards.
Defect One, Wall Thinning on the Outside of the Bend
How to identify it. Of all the tube bending defects, thinning is the hardest to see, because it is invisible from outside. Find it with an ultrasonic gauge along the outside of the bend, or by sectioning a first article and measuring with a ball micrometer. Take readings at the tangent points and at 45° and 90° through the arc, because the minimum is rarely at the midpoint. Report remaining wall in millimeters and as a percentage lost, since acceptance criteria are written both ways.
What causes it. The outer wall of a bend is longer than the tube it was cut from. Material on the outside is in tension along the tube axis, and metal in tension gets thinner. The tighter the radius, the greater the difference in path length between the outer fiber and the neutral axis, and the more the wall gives up. Thinning is geometry, present in every bend ever made.
What changes it on the equipment side is listed here in order of effect.
Opening the centerline radius is the largest lever and the cheapest. Going from 2D to 3D takes the estimate from 20% to 14.3% with no hardware change beyond a different bend die. If the drawing radius came from a styling decision and not a clearance constraint, that is the conversation to have before buying tooling.
A following pressure die is second. Driving the pressure die forward at or slightly above draw speed feeds material into the bend from behind instead of letting the machine pull it all out of the wall. On tight radii this decides whether the part passes, and it is a machine capability and not a tooling one. A bender without it cannot be given it with better dies.
Mandrel support matters less here than for the other two damage modes, but an undersized mandrel or one set too far back lets the section deform, concentrating strain and producing a thinner minimum than geometry predicts. Lubrication acts at the margin. Friction resists material feeding forward, so a dry mandrel converts feed into stretch.
Where the acceptance limit comes from. In pressure work it is not your choice. Power construction, boiler, bridge, shipbuilding and process piping codes all set a minimum remaining wall, and the correct procedure is to work the radius backwards from that limit and not to pick one and hope. Furniture, fitness equipment and decorative work carry no equivalent requirement, which is why the same machine is specified very differently for two customers.
Defect Two, Ovality and Flattening Across the Section
How to identify it. Take the maximum and minimum outside diameter at the same station with a micrometer or a ring gauge, then report the ovality as (Dmax − Dmin) ÷ Dnominal × 100.
Measure at several stations through the arc and not just at the crown. On a badly supported bend the section is widest across the plane of bending and narrowest through it, so a single reading in the wrong orientation looks acceptable. Where the tube must accept a fitting, a plug gauge or ball pushed through the bore tests the function directly.
What causes it. A tube resists bending as a closed section, and the moment it bends, the top and bottom walls are pulled toward the neutral axis. With nothing inside to push back, the section flattens. Because it is a stability problem, it scales with the wall factor. At a D/t of 10 the section largely holds itself, and at a D/t of 40 it will not hold itself at any useful radius.
What changes it on the equipment side. Ovality is the one of the tube bending defects that is most directly bought with tooling.
The mandrel is the primary control. A plug mandrel supports the section only at the tangent point and suits heavy wall at open radii. A linked ball mandrel supports it continuously through the arc, and the ball count should rise as the wall thins and the radius tightens. Under-specifying ball count is a common cause of ovality that survives every other adjustment. What a moving mandrel arrangement does, and where it is required, is set out on the mandrel tube bending page.
Die cavity fit is next and often overlooked. The bend, clamp and pressure die grooves form the outer half of the constraint. If the cavity is worn, or cut for a nominal OD the incoming tube does not actually hold, the tube has room to spread and will use it. Tube OD tolerance is a real input and not a formality.
Pressure die load closes the loop. Too little and the tube lifts out of the cavity. Too much and you trade ovality for drag marks and extra thinning. This is the first place to look when ovality appears mid-batch with no tooling change.
Soft materials deserve a note. Copper and aluminum are easier than mild steel on force and harder on shape. The constraint becomes marking and ovality and not tonnage, so clamp pressure and mandrel match matter more. Material behavior is treated in bending stainless steel tubing.
Defect Three, Wrinkling on the Inside of the Bend
How to identify it. Wrinkling is the only one of the tube bending defects that can be found without instruments. Ripples run around the tube on the inside of the bend, evenly spaced, starting at the tangent point where the bend begins. Lay a straight edge along the inside of the bend and light it from a low angle. Incipient wrinkling that a fingertip misses shows up at once as a shadow pattern. For production control, a physical go/no-go sample beats a written ripple height limit, because operators judge against an object more consistently than against a number.
What causes it. The inner wall of a bend is shorter than the tube it came from, so it is in compression. Compressed thin material can thicken uniformly or buckle, and buckling takes less energy, so unless something physically prevents it the wall buckles. Everything the tooling does against wrinkling amounts to removing the space a wrinkle needs.
What changes it on the equipment side.
The wiper die is the dedicated answer and the least forgiving item in the set. Its tip reaches into the gap between the bend die and the tube right at the tangent point, filling the space where a wrinkle would start. That tip must be fitted to a near-zero gap, which makes it a consumable and its condition a routine check. A worn wiper tip produces wrinkling that no pressure adjustment removes, and it is the first thing to inspect when a previously good job starts rippling.
Mandrel nose position is second, and the one most often set wrong. The nose should sit slightly forward of the line of tangency. Too far back leaves the tube unsupported exactly where the bend starts. Too far forward and the mandrel scores the bore and drives up drag. It is an adjustment on the scale of a millimeter with an outsized effect.
A following pressure die attacks the cause and not the symptom, because feeding material forward reduces the compressive load the inner wall has to absorb at all. Lubrication closes the list, and it matters more here than for thinning, because the wiper tip and mandrel both work in sliding contact under high pressure. Under-lubricated tooling galls, and galled tooling stops fitting.
Defect Four, Springback and Angle Error
How to identify it. Measure the included angle after the tooling releases, with a digital protractor or on a fixture, and compare it against the programmed angle. On multi-bend parts also check the plane of each bend and the distance between bend centers, because rotation and feed errors produce the same complaint, that the part will not fit, from different causes.
What causes it. Bending takes material past its yield point, but not all the deformation is plastic. When the dies release, the elastic portion recovers and the bend opens up. The size of the effect scales with the ratio of yield strength to elastic modulus, which is why material matters more than machine here. Aluminum has an elastic modulus around 69 GPa against roughly 200 GPa for steel, so at comparable yield strength it springs back on the order of three times as much. Austenitic stainless work-hardens as it forms, which raises springback and makes it vary between batches.
So springback is a material property and not a machine fault. Even a well built bender produces it. What separates machines is whether they compensate repeatably.
What changes it on the equipment side. Over-bend compensation is the whole answer to this last of the tube bending defects. The control bends past the target by a compensation angle, the tube relaxes, and the finished angle lands where it was wanted. On a CNC bender that value is stored with the program, established once on a first article and then repeated. On a manually set NC machine the same physics applies, but the value lives in the setter’s sheet, which works for long runs of one part and is awkward for frequent changeovers. That is the real distinction between the CNC pipe bending machine range and the NC semi-automatic tube bender range.
Mechanical headroom has to exist for compensation to be usable. A machine that stops at exactly 180° cannot over-bend a 180° return, so the maximum bend angle on a data sheet should always sit above the hardest angle you will be asked to make. Telhoo’s 219NC lists a 190° maximum bend for that reason.
Repeatability sets the floor on what compensation achieves. A value set on the first part is only valid on the thousandth if the bend, feed and rotation axes repeat. The honest limit is that on very heavy sections, variation between pipe batches shifts springback more than the machine’s own repeatability does. The answer is then incoming material control and a first-article check per batch, and not a tighter machine.
Draw speed plays a quieter role. Faster metal flow gives less predictable results, which is why bend rates fall as sections get heavier. Telhoo’s CNC 38 head runs up to 100°/s on small tube, while the 130NC is run deliberately slowly at 12°/s and the 168NC at 4°/s on Φ168 × 14 mm. That is a quality decision and not a hydraulic limitation.
Fixing One Defect Often Creates Another
This is what separates a bending cell that converges on a good part from one that oscillates, and it is the reason a change to one setting should always be followed by a fresh measurement of all four defects, not only the one you were chasing. Most adjustments push thinning and wrinkling in opposite directions, because they act on opposite walls of the same bend.
| Adjustment | Helps | Risks |
|---|---|---|
| Increase pressure die load | Ovality, wrinkling | Drag marks, slightly more thinning |
| Increase following pressure die feed | Thinning, wrinkling | Too much piles material up at the tangent |
| Move mandrel nose forward | Wrinkling, ovality | Bore scoring, higher draw load, mandrel wear |
| Add mandrel balls | Ovality on thin wall | More drag and lubrication, harder extraction |
| Open the centerline radius | All four defects | New bend die, and the part geometry may not allow it |
| Reduce draw speed | Springback consistency, wrinkling | Cycle time |
| Increase over-bend angle | Springback only | Nothing, given headroom past the target |
One row stands out. Opening the radius is the only adjustment that improves all four defects at once, and the only one whose cost is paid in tooling and drawing changes and not in a new problem.
The row worth a warning is the following pressure die feed. It is the most powerful setting for thin wall work, and overdriving it piles material up at the tangent in a way that looks like the wrinkling it was meant to cure. If wrinkling worsens as the feed goes up, the feed is now the cause.
The Order to Troubleshoot Tube Bending Defects In
Defects are usually chased in the wrong sequence, starting with the settings easiest to change and not the ones most likely to be responsible. This order is more productive, and the right column gives the reason each step comes where it does.
| Step | Do this | Why it comes here |
|---|---|---|
| 1 | Measure remaining wall, ovality percentage, ripple condition and finished angle | Most bending arguments are two people naming different defects with one word |
| 2 | Calculate CLR/OD and D/t against the drawing | 1.5D at a wall factor of 40 is out of envelope and not misadjusted |
| 3 | Verify the tube, meaning actual OD, actual wall, weld position and hardness | Tube at the low end of wall tolerance turns a proven job marginal |
| 4 | Inspect wiper tip, die cavity, mandrel balls and mandrel diameter | Worn tooling produces defects that no setting corrects |
| 5 | Set mandrel position, then pressure, then following feed, one at a time | Moving two at once leaves a compensating pair of wrong values |
| 6 | Tune springback compensation | It masks nothing, so doing it earlier just means doing it twice |
Step 3 explains most defects that appear overnight with nothing changed in the cell, and the weld seam position relative to the bend plane belongs in the setup sheet and not in tribal knowledge.
Trade coverage of forming problems and setup practice is collected in the tube and pipe fabrication resources at The Fabricator, and the code framework that sets remaining wall and ovality limits for pressure work is published through ASME Codes and Standards.
What This Means When You Specify a Machine
Knowing the tube bending defects is the honest way to read a bender specification, because it turns numbers into questions about your own parts.
Ask about the pressure die assist and not just tonnage. Bending force decides whether the machine can form your heaviest section at all, and the following pressure die decides whether the result passes on your tightest radius. Only one of the two appears in a headline capacity figure. Read the rated wall the same way. A machine rated for Φ219 × 20 mm is not a small machine scaled up, because at the larger size the limiting problem changes from buckling to force.
Check mandrel length against your part and not your tube. The maximum mandrel length, which runs from 2,250 mm on the 50NC up to 6,000 mm on the 168CNC across Telhoo’s range, governs how far into a tube a mandrel-supported bend can be placed. A part whose last bend sits beyond that reach is formed unsupported and will show ovality whatever else is right.
Match the control to your changeover frequency. Where springback compensation must be re-established several times a shift, storing it in a program is worth real money. Where one part runs for a week, it is not. That is the dividing line between the CNC and NC ranges, and it is set out alongside the forming methods on our tube bending solutions and rotary draw bending pages.
If you have a drawing with a radius you are not sure is achievable, the fastest route to an answer is a sample. Send the tube specification, the wall thickness, the radius and the governing code if there is one, and tell us what you need to bend. Our engineers will say whether the geometry works before anyone quotes tooling for it.
