Well, whether the material stretches or compresses on either side of the neutral axis, while the volume may change EXTREMELY slightly (the metal compressing on the inside of the neutral axis etc. - might not even be measurable), in REAL life the MASS itself doesn't change. HOWEVER, since Creo (or any other CAD program that I know of) does NOT thin the model out in the bend area, I would think that when it adds a bend, there is a slight volume increase in the bent area, and when a form is used the material is not stretched and thinned, and that's probably what is throwing the mass calculation after the bends off. Honestly, since the flat pattern is usually created AFTER the bent shape, which is more accurate? Hmmm... I think if the mass is critical (i.e. NASA work like I was doing), it's probably best to actually make a part that conforms to print, weigh it, then assign a fixed mass to it, and that shouldn't change whether it's a flat state or not.
I'm not an expert in sheet metal though, so...
