A whole series could be developed on different ways to fail weld qualifications using bend tests.
Let’s start with one way that bend tests can differ. For the same required elongation, bend tests can be made in different widths. For instance, the side bends from a thick part can vary by a factor of two in width in standards like AWS B4.0.
The choice of thickness is not listed in the standard as connected to the relationship between outer surface elongation and specimen bend radius and thickness.
But the choice of width does have an effect on the stress in the direction across the width. This is can be described in terms of fancy concepts like constraint or transition to plane strain. But it can also just be described by saying that these stresses in the cross direction rise with width up to about a width of 8 times the specimen thickness. Most bend test specimens have smaller widths, perhaps as low as two times the specimen thickness.
The higher stresses in the cross direction can induce flaws that are pooly oriented for failure due to the longitudinal stress to fail under the cross direction stress. This will tend to happen near the middle of the specimen width. So more failures can happen.
EWI can help you with bend tests, for instance, by performing them, modeling them, assessing failures, and recognizing options. You can call me, Bill Mohr, at 614-688-5182 to get connected with these resources at EWI.