Springback is the most common reason a bend measures wrong when the machine did exactly what it was told. The program said 90 degrees, the machine drove the die through 90 degrees, and the part came out at 87 or 88. Nothing was faulty. The tube simply recovered. Springback is a property of the material and not a fault of the machine, and a good bending process treats it as a number to be measured and managed.
This article explains why it happens, what moves it, how to prove a compensation value and where to store it. It builds on the short treatments in pipe bending calculations and tube bending defects and what causes each one, and it adds the working procedure and the machine side. Springback compensation tube bending is also where the difference between NC and CNC control shows most clearly.
Why a Bent Tube Opens When the Dies Release
Bending takes the material past its yield point on the outside and the inside of the bend. Near the neutral axis, though, the strain is small, and in the whole section some of the deformation is still elastic. While the dies hold the tube, the elastic part is held in place. When the clamp opens and the pressure die backs off, the stored elastic strain is released and the tube relaxes toward straight. The angle gets smaller, and the radius gets a little larger, because the arc opens.
The amount depends on how much of the strain was elastic. A material with a high yield strength and a low elastic modulus stores more elastic strain and springs back further. Aluminum has a modulus roughly a third of steel, so at comparable strength it springs back more. Stainless hardens as it is bent and springs back more than carbon steel. Annealed copper has a low yield strength and springs back little in degrees. The material comparison is in aluminum tube bending and how tube metals compare and the stainless details in bending stainless steel tubing and what changes.
What Moves Springback Compensation Tube Bending Up or Down
Six things change the amount, and each matters when a correction is carried from one job to another.
| Factor | Effect on springback | Why |
|---|---|---|
| Material and grade | Large | Yield strength and modulus set the stored elastic strain |
| Temper or condition | Large | Harder tempers store more elastic strain |
| Radius ratio | Large, in a surprising direction | A tighter bend puts more of the wall into plastic strain, so less is elastic |
| Wall factor | Moderate | A thin wall behaves more elastically and springs back more |
| Lubrication and drag | Moderate | Drag changes where the material yields |
| Bend angle | Roughly proportional | A longer arc stores more elastic strain in total |
The radius effect surprises people. A tight bend at 1.5D springs back fewer degrees than a generous bend at 4D, because more of the section has been pushed into plastic flow. Tight bends are difficult because of wall thinning and wrinkling and not because of angle. UK and Australian drawings call the radius the centreline radius, and the ratio is defined in centerline radius explained for tube bending.
The lot matters as much as the grade. Two coils of the same nominal grade can differ in yield strength, and the difference shows up as a different springback. That is why a correction proven on one lot is a good start for the next and not a guarantee. A change of supplier, temper or mill can move the value enough to need a new proof.
How Compensation Works in Practice
The principle is simple. The control bends past the target by the amount of the springback, the tube relaxes, and the finished angle lands on the target. The programmed angle equals the target angle plus the springback. What takes care is finding the springback, and the way to find it is by measurement and not by calculation.
Here is a worked example for illustration, using assumed numbers. A part needs a 90° bend. A trial piece is bent with a programmed angle of 90°, unclamped and handled, and then measured cold at 87.5°. The springback is 2.5°. The programmed angle for the next piece is 92.5°. The second piece is bent, measured and found at 90.1°, which is within tolerance, so the value is confirmed. A third piece is checked a little later in the run to confirm that it holds.
The example also shows the scaling. Springback is roughly proportional to the bend angle, so a 45° bend on the same tooling and material would be expected to spring back about 1.25°, and the programmed angle would start at 46.25°. That is a starting estimate and not a rule, because the radius ratio and the strain pattern change with angle. The estimate is checked on a trial piece, and the corrected value is stored for that bend.
| Step | What to do | Example value |
|---|---|---|
| 1 | Bend a trial piece at the target angle | Programmed 90° |
| 2 | Unclamp, handle and measure the angle cold | Measured 87.5° |
| 3 | Work out the springback | 2.5° |
| 4 | Add it to the programmed angle | Programmed 92.5° |
| 5 | Bend a second piece and measure | Measured 90.1° |
| 6 | Store the value and check again later in the run | Stored 92.5° for this bend |
Measuring the Released Angle
The measurement decides the quality of the correction. Measure the angle after the part has been unclamped and handled, because the part can relax a little further when it is picked up, and measure it cold. A part measured while still in the machine, or while warm from a long bend, will give a different answer than the one the customer sees. Use the same method every time, and use a method that is as accurate as the tolerance requires.
The usual tools are a digital protractor or angle gauge, a template or checking fixture, and for tight tolerances a measuring arm or coordinate measuring machine. A fixture is often best on production parts, because it measures what matters, which is whether the part fits, and not an angle in isolation. On parts with several bends, check the plane of each bend and the distance between bend centers as well, because rotation and feed errors can look like angle errors.
Take more than one piece. Springback varies a little from piece to piece, and the value to use is a representative one and not a single reading. Three trial pieces followed by two confirmation pieces is a common routine, and it also shows how consistent the material is. If the readings scatter widely, the problem is probably the tube or the lubrication and not the compensation.
Compensation on NC and CNC Machines
The compensation is the same physics on every machine, and what differs is where it lives. On a CNC bender the over-bend is stored in the program against each bend, so once it is proven it is applied automatically on every part and every later run. Telhoo’s published CNC range lists bending precision of ±0.1° and a stored program for each part, and its 63CNC case describes the control storing hundreds of tube programs. A new operator can then run a proven program without re-establishing the correction.
On an NC semi-automatic bender the angle is controlled digitally, with repeatability published at ±0.1°, and the setter enters the over-bend as part of the setup. The value lives in the setup sheet and in the setter’s experience. That works well for long runs of one part, and it is more demanding for frequent changeovers, because each new setup needs the value found or retrieved. The comparison is in CNC vs NC pipe bender and how to choose.
In both cases, the machine must be able to reach the over-bend. A machine that stops at exactly the target angle cannot compensate. Telhoo publishes a maximum bend angle of 190° on its CNC range and on models such as the 219NC, and 185° on several NC models, and a 180° return bend can be over-bent by a few degrees. Check the maximum angle against the hardest bend you will make plus the springback.
A Second Example, Same Part and a New Lot of Stainless
This example is illustrative and uses assumed numbers. A stainless part has run for months with a stored over-bend of 3.0° on a 90° bend. A new lot of tube arrives from a different mill with the same nominal grade. The first piece is bent with the stored value, the programmed angle of 93.0°, and measured cold at 89.2°. The part is 0.8° short, which means the new lot springs back 3.8° and not 3.0°.
The correction is updated to 3.8°, the programmed angle becomes 93.8°, and two more pieces are bent and measured, both inside tolerance. The new value is stored against the new lot, and the old value stays on file for the old lot. The check took a few pieces and a few minutes, and it prevented a batch of short parts. That is the case for a routine first-piece check on every new lot, and for storing the correction against the lot and not only the part number.
| Lot | Programmed | Measured cold | Springback | Stored value |
|---|---|---|---|---|
| Lot A, proven | 93.0° | 90.0° | 3.0° | 3.0° |
| Lot B, first piece | 93.0° | 89.2° | 3.8° | Update needed |
| Lot B, confirmed | 93.8° | 90.0° | 3.8° | 3.8° |
Springback by Material, a Practical Guide
The materials rank in a consistent order, and the ranking helps plan the number of trial pieces. Annealed copper springs back least in degrees, but it ovalizes and marks, and the angle is rarely the main worry. Low-carbon steel springs back moderately and steadily, and a proven value holds well from lot to lot. Austenitic stainless springs back more, rises with work hardening, and varies more between grades and lots. Aluminum springs back the most, depends strongly on temper, and a value proven on one temper does not hold on another.
That ranking suggests where to spend effort. For steel, a first piece check on each new lot is usually enough. For stainless, allow more trial pieces and store the value per lot. For aluminum, record the temper with the value and treat a temper change as a new job. And for any material, treat a change of supplier as a new lot. The same ranking is behind the materials article, and it shows why mixed-material shops store a program per material.
The Radius Also Changes
Springback is usually described as an angle error, and it is also a radius error. As the tube relaxes the arc opens, so the finished centerline radius is a little larger than the die radius. For most parts the effect is small enough to ignore, because the tolerance is on the angle and the end points. On parts where the radius itself is critical, such as a bend that must clear a nearby surface, check the finished radius on the trial piece and not only the angle.
The change in radius also changes the arc length slightly and the position of the next bend. That is another reason the correction is applied per bend. The arithmetic of arc length, setback and deduction assumes the die radius and the programmed angle, and the springback adjusts the result a little. The cut length is established on the first article and not only calculated, as set out in the calculations article.
Fixtures, Templates and Gauges
The method of measuring should match how the part is judged. A digital protractor is quick and accurate for a single angle in a plane. A template, a profile that the part must lie in, checks the whole shape at once and is easy for operators to use. A checking fixture with locating pins and gauges on the ends checks the end points, which is what an assembly cares about. A measuring arm or a coordinate measuring machine checks any feature and gives numbers for the record.
Choose the simplest method that answers the question, and use it consistently. A fixture is often the best for production, because it shows whether the part will fit and not only whether one angle is right. On the first article use a precise method to establish the correction, and then use the fixture to confirm and to monitor through the run. Keep the gauges and fixtures checked, since a worn gauge makes a good part look bad and a bad part look good.
What to Record for Each Compensation Value
A correction is only useful if it can be found and used later. Record the part number, the bend number and angle, the material, grade and temper, the lot or heat, the tube outside diameter and wall, the tooling set and the lubricant, the programmed and measured angles, and the date. On a CNC machine most of this lives with the program. On an NC machine it belongs in the setup sheet, and the sheet should be kept with the tooling for that part.
The record serves three purposes. It lets the next run start from a proven value. It lets a change in the result be traced to a change in the inputs, such as a new lot or lubricant. And it is evidence for customers who audit process control. A shop that can show the correction, the lot and the check on each batch can answer most questions about angle accuracy without a search.
Troubleshooting a Value That Will Not Settle
If the angle varies from piece to piece after the correction is set, the cause is usually something other than the correction. The table lists the usual suspects.
| Symptom | Likely cause | What to check |
|---|---|---|
| Angles scatter widely | Clamp slip or tube movement | Clamp length, clamp force, tube surface |
| Drift through the run | Warming oil or tooling wear | Oil temperature, die and mandrel condition |
| Step change at a new lot | Different yield strength or temper | Mill certificate and a first piece check |
| Short angle on every piece | Correction too small | Remeasure cold and recompute |
| Correct angle, wrong end point | Feed or rotation error | Length and rotation settings |
| Angle varies with lubrication | Drag changes the strain pattern | Lubricant type and quantity |
Fix the cause before changing the correction. Adding more over-bend to cover a clamp that slips makes the average right and the scatter worse, and it hides the real fault.
Thin Wall and Heavy Wall Compared
Wall thickness changes both the amount of springback and how reliably it can be controlled. A thin wall relative to its diameter carries less material in bending and behaves more elastically, so it tends to spring back more, and it also ovalizes more easily, which blurs the measurement of the angle. A heavy wall works more plastically and springs back less in proportion, but it takes more force and releases more stored energy in absolute terms, and the machine frame itself can flex and add an error that looks like springback.
That last point is why heavy-wall benders are built rigid and run slowly. Telhoo’s pages describe the 168NC bending at 4°/s and the 168CNC at 8°/s on Φ168 mm pipe, and the 168CNC page notes that slower flow gives more predictable springback. A faster bend can change how the material yields, and the correction found at one speed may not hold at another, so keep the speed the same between the proving piece and production.
Machine Features That Help
Several machine features make compensation easier. A maximum bend angle comfortably above the largest target gives room to over-bend. A control that stores a correction per bend and per part number removes the need to remember it. Repeatable axes, with feed and rotation as well as angle held to a small fraction of a degree or a millimeter, mean the value found on the first piece is still valid on the thousandth. And a rigid frame and stable clamping mean that the difference between pieces comes from the material and not from the machine.
Ask a supplier for these specifics when comparing machines, and read them against your parts. The published repeatability of Telhoo’s CNC range is ±0.1° on bending and rotation and ±0.1 mm on feed, and the NC range publishes ±0.1° on bend angle. They are the numbers that decide how far a proven correction holds across a run, and the selection sequence is in how to choose a tube bender from the part drawing.
Bends Close Together and Multi-Bend Parts
On a part with several bends, springback in one bend can shift the position of the next. If the first bend relaxes by 2.5° and that is not compensated, every later feature is rotated by that amount about the first bend, and the error grows along the part. The correction therefore has to be applied per bend and not per part, and the bends are compensated in sequence. On a CNC machine this is routine. On a hand-set machine it is a matter of discipline and a good setup sheet.
Bends close together raise another point. A bend made next to an earlier bend works on material that has already been strained, and the springback can differ from an isolated bend. Where two bends are within a few tube diameters of each other, prove the second bend in place on a real part and not on a straight piece. The tolerance stack across the whole part is covered in automotive tube bending by part family and volume.
Common Mistakes With Compensation
The same mistakes recur. The first is measuring while the part is warm or still in the machine. The second is trusting a table value, because tables describe typical material and not your lot. The third is applying a value found at one angle to another without scaling and checking. The fourth is forgetting that a change of lubricant, tooling or supplier can move the number. And the fifth is correcting the cut length instead of the angle, which leaves the angle wrong and the length wrong in a different way.
A sixth mistake is chasing a springback problem that is really a tooling problem. If the angle varies widely from piece to piece, check the clamp for slip, the pressure die force and the lubrication before changing the over-bend. Slip gives a short bend that looks like springback and does not behave like it, and the clamp and tooling checks are in tube bending tooling and how to match a die set to the job.
What to Send for Springback Compensation Tube Bending
- Material, grade and temper, and whether the lot may change between batches.
- Tube outside diameter and wall, with the tolerance you can hold in purchasing.
- Every centerline radius and every bend angle, including the largest angle on the part.
- The angle tolerance and how it is checked, whether by fixture, template or measuring arm.
- Bends per part and the spacing between bends.
- Batch size and variants, since they decide how much a stored correction is worth.
Send those and the reply can say which machine, control and tooling fit, and how many trial pieces to plan. Tell us about the part, and see the two ranges on the CNC automatic pipe bending machine and NC semi-automatic tube bender pages. The wider process view is on our tube bending solutions page.
