Ive had this happen on occasion, and it was the particular jig I used and I learnend that it would slide off center as the plunger went into the part. I just off set my piece as I put it in and they come out fine when I'm done.
I have had where I had to take a second slice out of the remaining center drop to re-test a slice that didn't bend correctly.
It's not the welders fault if QC messed up the bend test, Our coupons are 6" long not 5" so I have a little extra to take it from.
It's getting to be very aggravating. I'm using a guided bend, not a wrap around. I think a wrap around may work better so one side could be clamped in. I'm assuming the bend is taking the least path of resistance which seems to slide to the square edge until it looks like a candy cane. We are having to offset like you said but its a hit and miss with the many students we have doing this test.
got a picture of your setup?
Are both pieces of plate the same material ? I have had to offset P1-P5 ASME PQR bends. Can't remember to which side.
Is your "candy cane" going to the same side of the jig each time?
Is it a roller jig or with a female die. If a doe, check for burrs flat spots etc. If a roller, look at the shaft/bearing/axle on both closely. They may spin free with no load but act differently with a few tons of force.
Have a nice day.
Gerald Austin
I know this an old post but, if you have enough material you could go with longitudinal bends instead. It takes a much longer test plate but, that's what we do with materials of dissimilar strength.
Plunger and die jigs that are loose or not true (parallel and in line) to each other are bad about "candy caning".