By OBEWAN 
Date 09-02-2013 21:31
Edited 09-02-2013 23:05
You need to post some more details:
What is the base material, joint design, process, and code? Are we talking about a fillet weld or a flare bevel weld? Are you using a WPS? Do you have a qualified repair procedure? I presume we are talking about a fillet weld, and most likely the repair would involve grinding out the crack to solid material and rewelding with the same original process. Some codes restrict the number of rewelds that are allowed.
It is likely that it can be successfully repaired. If it cracked from the weld crater there may have been a minute defect in the weld termination that passed visual inspection. Perhaps a DPI or FPI method would have revealed the defect. But, there is no way to tell how much stress it absorbed.
Hello GGordon, Fabshield 21B would not be my first choice for welding any items that might be subjected to much in the way of impacts or other cyclic stresses. If you check the description of this particular filler wire you will probably see that it doesn't qualify for seismic application. This essentially, in a short description, means that it isn't very ductile and will crack long before it generally bends or deforms. Charpy values for this filler are very low and this is one of the reasons that it is not allowed for seismic applications, I believe that it also is limited in multi-pass applications or bead size. I am sure that others will reply and likely give a more detailed and accurate description than I have here, yet, the answers will still likely suggest that the issue lies with the filler wire selection and how it was stressed in the straightening process. Good luck and best regards, Allan
It would be good to know the size of the welds, both leg size and length, as well as all the other info Obewan asked for.
I would agree with Allan's assessment of the weld electrode used. Probably not the best for the beating it took.
Have a Great Day, Brent
Keep in mind that CVN's are not necessarily designed to determine brittleness of a material as such. Bends are designed for this. CVN's are designed to determine the ductile to brittle transition in the context of a fracture rate. If the numbers are low this is an indication that the material is below DBTT. If the numbers are good then it is generally considered above. Even materials that have schytty CVN's can have exceptional toughness above the DBTT.
In this case you have a high impact rate, however, you may not be below the DBTT of the material in which case the CVN values may be of little value.
Are you cleaning off the Mill scale? With a small wire and that low of voltage you might just be laying a weld on the surface with only the center making decent penetration.
Jordan
