September 26, 2002
Bend tests for aluminum are different than for steel. For instance, lack of fusion discontinuities in aluminum usually don't show up as well in radiographs. A bend test is a more discriminating way to test your aluminum weld.
Almost all welding codes require bend tests for welding procedures and welder qualifications. These tests can cause more problems in aluminum than in steel.
In bend testing, a 1 1/2-inch-wide sample is cut so the weld is running across the width of the sample. Then the sample is bent around a specified radius, with either the root or the face of the weld in tension. The bend usually is made using a radiused plunger that pushes the sample down into a radiused female die.
Sometimes I make aluminum bend test samples that bend only 30 or 40 degrees and then snap. What's going on?
This isn't uncommon in aluminum bend tests. A sample that has broken this way is shown in Figure 1. The sample started to bend a little bit in the die and then snapped abruptly.
The reason it snapped is shown in Figure 2. The dark gray line in the lower left part of the photo is the remnant of a fine defect in the weld caused by lack of fusion.
This is the first thing you should look for in a failed bend sample. Any lack of fusion in the weld, no matter how small, will make the sample snap on bending. When a sample snaps, don't blame the welding wire or the parent material; a defect almost certainly is in the weld.
But wait. You radiographed the test plate before you cut the bend samples from it. Why didn't you see the lack of fusion on the radiograph?
Lack-of-fusion discontinuities in aluminum usually are tight and don't show up well on radiographs. It's fairly common to miss lack of fusion in a radiograph of an aluminum weld. The bend test is more discriminating.
I don't see any lack of fusion in the bend test sample, but it still fails. Why?
Aluminum alloys are different from steels in several ways. One of these ways is that aluminum alloys and aluminum welds aren't as ductile as steels. For this reason, the bend radius required for aluminum alloy bend tests usually is larger than that required for steel samples.
Don't guess. Check the code you will use in production. You often will find that you're bending samples around a too-small radius, which almost guarantees failure.
A special case allows for the limited ductility of welds in 6000 series alloys, such as 6061-T6. If these bend samples are machined to the standard thickness of 3/8 in., they often fail.
The major welding codes recognize this. American Society of Mechanical Engineers (ASME) Section IX, while not explicit, allows you to machine 6061 or 6063 samples down to 1/8 in. thick for testing (QW-466.1). American Welding Society D1.2 specifically states that welds in M23 materials (6000 series alloys) should either be tested as-welded, with the bend sample machined to 1/8 in. thick, or annealed before testing with the sample machined to 3/8 in. thick (paragraph 184.108.40.206). Many fabricators and welders are unaware of these exceptions and continue to fail procedure and performance qualifications needlessly.
The other major difference between welds in aluminum alloys and those in steels is that a weld in steels normally is about the same strength as the surrounding material, while in aluminum alloys, the weld and the adjacent heat-affected zone (HAZ) usually are much weaker than the surrounding material.
When a relatively homogeneous steel sample is bent in a plunger-type bend tester, the sample bends smoothly. If an aluminum sample is bent in the same tester, the relatively weak HAZ bends sharply, with little or no bending in the rest of the sample. The HAZ then kinks and finally fractures at the crack before the rest of the sample deforms at all.
The way to prevent these failures is not to use a plunger-type tester for aluminum samples. Both D1.2 and Section IX allow the use of wraparound bend testers. In fact, the 2002 version of D1.2 specifically will state that the wraparound tester is the preferred type for aluminum bend tests.
For those who aren't familiar with this type of bender, it's similar to a small rotary draw bender used to bend tubing. In use, one end of the sample is restrained. The swing arm then pulls the sample around a die with the proper radius. These testers are shown schematically in Figure 4.15 of D1.2 and in QW-46.3 of Section IX.
A few additional factors can cause bend samples to fail needlessly. All codes allow the four corners of the bend sample to be radiused. Take the time to perform this extra operation. Sharp corners often start cracks that will cause failure.
It's also important to take care to get a smooth surface finish on the sample. The weld reinforcement must be removed completely by machining. Even a small notch at the weld toe (on a face bend) or at the edge of the weld root can cause bend test failure.
Even though it might sound like overkill, make sure any machining marks are along the length of the sample. Believe it or not, machining marks that go across the width of the sample can act as starting points for cracking during testing.