Testing standard
Metallic materials — Bend test
Written and technically reviewed by Dak System Inc. engineeringLast reviewed
ISO 7438 is the bend test for metallic materials. A piece is bent in one continuous movement to a stated angle, to parallel legs, or to legs in contact, and the outside of the bend is then inspected. Nothing is plotted and nothing is calculated — the output is the condition of the convex surface, judged against what the product standard demands.
From the test method to your testing system
Explore the DAK machines already listed for ISO 7438, then review the grips, measurement and setup requirements below.
Universal Testing MachineSeries 7200Explore the machine →
Universal Testing MachineSeries 9000Explore the machine →01Understand the method
A piece of metal is bent, in one continuous movement, until it reaches a prescribed condition — a stated bend angle, legs parallel at a set spacing, or legs in contact. Three devices are recognised: the commonest sits the test piece across two supports and drives a former into the middle; the second replaces the supports with a V-block, so the piece folds into the vee as the former descends; the third clamps one end and wraps the free length round a former. The load is then removed and the outside of the bend inspected.
Nothing is plotted and nothing is calculated. The output is the state of the convex surface after bending: either it meets what the product standard demands, or — where that standard is silent — it is judged on the absence of visible macro-cracks.
That binary answer interrogates the surface and near-surface structure, where forming operations actually fail. A plate that passes a tensile test comfortably can still split on the brake if it carries rolled-in laps, a decarburised skin or heavy segregation bands: the tensile coupon shares strain along a gauge length, while the bend concentrates it into the outer fibre. Bend results are an acceptance gate for delivery condition, forming behaviour and heat treatment, not design data.
02Prepare the specimen and test settings
Three bending devices are recognised, and the product standard chooses between them along with the former diameter and the required condition.
03Build the test setup on a DAK machine
The standard has nothing to say about the frame. Every dimensional clause governs the jig; the press behind it is the laboratory's choice, sized to the heaviest section it must fold. Stroke deserves the closer look: bending to parallel or contacting legs eats far more travel than bending to a stated angle.
Geometry is what the standard controls. Support span is an output, not a setting — l = (D + 3a) ± a/2, from former diameter and test-piece thickness — and it must hold through the bend. Set it shorter and the supports hold the ends instead: the piece is drawn rather than folded. V-block edges are radiused in proportion to specimen thickness. Former diameter is the product standard's to set, not the laboratory's, and the nearest one to hand is a different test.
The bending force is applied slowly, so the material flows plastically rather than being snapped round the former, and clause 6.3 gives the figure that settles the argument: in case of dispute, a testing rate of (1 ± 0,2) mm/s — 60 mm/min at the former — shall be used. Set that rate whenever a result may be contested. No force accuracy class is prescribed: the method controls fixture geometry, not frame metrology, so a load reading is a process indication rather than a result. No extensometer is used.
The adjustable-span beam gives the supports-and-former arrangement, with the span and former set to what the product standard calls for. No curve is recorded here — the fixture only has to deliver the geometry accurately.
SpecificationsDak verifies against whichever standard the method names, and where a class applies our frames sit a class tighter than it asks.
| The method asks for | Dak supplies | |
|---|---|---|
| Capacity | As with any bend test the force is set by the section and the span rather than by the method: thin sheet needs only a few kN, while thick plate and heavy bar coupons can call for several hundred kN, so bend jigs are commonly fitted to frames in the 300 kN to 600 kN class. | Load cells from 1 kg to 60 ton on the Series 7200, and 0.5 to 100 kN on the Series 9000 |
| Gripping | Three-point device with two supports and a former, or a V-block and former, or a clamp-and-former device | Our bend fixtures, built to the specimen |
| Environment | Ambient 10 °C to 35 °C in general, and 23 °C ± 5 °C where controlled conditions are called for | 3009 series chambers, −150 °C to +400 °C — temperature only |
04Run the test
05Report and interpret
Span drift is the classic. Rollers free to move, or supports set closer than the geometry allows, clamp the ends and quietly change the deformation mode; the piece looks bent and the strain history is wrong.
Former misalignment is the second: a former not centred between the supports produces an asymmetric bend, and cracks appear off the apex where nobody is looking.
Edge condition is the third and the most argued over. Rounding the edges is a requirement, not good practice, and a piece arriving at the jig square-cut or flame-cut has not met the method; what it then does at the bend is not evidence about the steel. Related: where thickness has been reduced by machining, putting the machined side in tension tests a surface the product does not have, and usually flatters it.
06Compare methods and find answers
| ISO 7438 | ASTM E290 | |
|---|---|---|
| Devices | Supports-and-former, V-block, wrap | Guided, semi-guided, free, bend-and-flatten |
| Output | Surface condition | Surface condition |
| Acceptance | From the product standard | From the product specification |
| Calculated quantity | None | None |
Both are mechanics rather than acceptance standards. A bend result travels only with the product standard it was judged against and the former diameter it was bent around — without those two it is not a result at all.
It is the international standard for the bend test on metallic materials. A piece is bent in one continuous movement to a stated angle, to parallel legs or to legs in contact, and the outside of the bend is inspected afterwards. The result is the condition of that surface — there is no number and no curve.
It interrogates the surface and near-surface structure, which is where forming operations actually fail. A plate that passes tension comfortably can still split on the press brake if it carries rolled-in laps, a decarburised skin or heavy segregation bands — a tensile coupon shares strain along its gauge length, while a bend concentrates it into the outer fibre.
Whatever the product standard demands. ISO 7438 supplies only the mechanics of making the bend. Where the product standard is silent, the judgement is the absence of visible macro-cracks on the convex surface. A bend result quoted without naming the product standard it was judged against cannot be verified.
Because it sets the strain in the outer fibre. A tighter former means more strain and a more severe test, so the same material can pass around one former and fail around a smaller one — both correctly. The former diameter is therefore part of the result rather than a setup detail.
The one the product standard names. The supports-and-former arrangement is commonest; the V-block folds the piece into a vee as the former descends; the wrap device clamps one end and bends the free length round a former. They impose different constraints, so they are not interchangeable choices for an operator to make.
No. It is an acceptance gate for delivery condition, forming behaviour and heat treatment. It answers whether the material will survive a specific imposed strain at the surface, which is a manufacturing question — not a quantity that goes into a stress calculation.
Discuss your specimen, test requirements and reporting needs with DAK engineering.
This page describes the method as practised. The governing text is the current edition from the issuing body. Tell us what you are testing and we will answer with the machine, the fixture and a quotation.
