This is just one more case where the actual discontinuity is not described properly due to the use of nonstandard terminology.

I defer to the standard terminology contained in AWS standards such as A3.0 Standard Welding Terms and Definitions or B1.11 Guide to Visual Examination of Welds. That makes sense to me considering I am in the United States and I am working with AWS standards.

When dealing with fillet welds, there is no "penetration" required, i.e., it is required the weld have fusion to the root, but not beyond to meet the requirements of AWS D1.1's performance test for fillet welds. When the fillet weld does not fuse to the root, it is a case of incomplete fusion, not incomplete penetration.

Joint penetration is the term used to describe a weld where the fusion extends into the joint. I differentiate between groove welds where the weld is in the joint and fillet welds are on the joint (my definition). 

When limiting the discussion to fillet welds, fusion to the root is not usually verified by visual examination alone. One method of verifying fusion to the root of the fillet weld is to section the weld transverse to the weld's axis and polish and etch the cross section to see if there is fusion to the root. Another is to grind the end of the fillet weld to see if there is fusion in the root. In either case, at least a portion of the weld is sacrificed.

Incomplete fusion can be present in other areas of the fillet weld as well as in the root. There may be incomplete fusion at the toe of the fillet weld. We usually refer to that condition as overlap. We can have incomplete fusion at the fusion face where the weld simply does not fuse to the fusion face or simply put, it doesn't fuse to the metal surface because of the presence of oxides, insufficient heat input, etc. In the case of a multi-pass weld, there can be incomplete fusion between the weld beads or the weld layers because of poor weld bead profile, etc.

I hope I answered your question. If not, someone (John, Brent, etc.) will chime in and give us their explanation. Between all the responses, one or more should clarify the situation for you.  

Best regards – Al

I would only add this to Al's very informative, accurate, and detailed response...

You started by asking rather the concave weld profile was acceptable (my paraphrase).  To which I say that it is as long as it meets the measurements required by the engineer.  If he/she called for a 5/16" fillet weld we would usually say it would be a convex weld with legs at least 5/16" long measured from the joint to the shortest toe of the weld.  But, if the welder happened to get a concave fillet instead of a convex then we use the other part of our gauge to make sure the throat is of the proper dimension to equal the strength called for by the 5/16" fillet weld. 

So, if the weld in question had a throat of the proper dimension to qualify as acceptable to the job specs then that in itself is not a factor in the failure.  Now, if the engineer underdesigned the weld then it was not large enough but that is not the welder's nor the inspector's problem.  That is an easy check to verify.

If the problem was lack of fusion/penetration into the parent metals that is easily determined as well.  Did it rip or just peel off with one surface having no indications of fusion?  What welding process was used?  Did someone try to use short arc where it should not have been used?  How about weld position?  Was it welded vertical DOWN?  That would also have resulted in a concave profile more than a convex one.  And that would also account for a 'lack of fusion' or better, incomplete joint penetration.  Welding downhill is often done fast and hot which but not necessarily deep penetrating.

There are lots of ways to go if someone is trying to affix blame or make excuses as to it wasn't their fault.  But the fact that it was concave instead of convex in profile is not an indication of a discontinuity that would have precipitated a failure.  Only that it needed to be thick enough through the throat to give the engineered strength. 

A GUESS based upon the very limited information presented: someone welded vertical down around the pipe and while the legs were long enough the throat wasn't so we have a double issue- undersized weld with incomplete joint penetration.  A formula for failure.  But the profile of the weld was not the problem.

Have a Great Day,  Brent

Most people do small welds with GTAW and have too small of a tungsten and to low output of a machine.  It will not adequately weld the type of application referenced here, unless, as Al stated in a later post, it is on some pretty thin tubing.  But then, if it was that light weight, why would it have broke.  They also are running very small beads that don't carry the kind of heat for depth of fusion since they are running small tungsten.

Another example, GMAW-S, (you asked about my statement which was not solely directed at this threads application).  You are not to use it as a Pre-Qualified on structural and for good reason, it is not adequate for a large percentage of the welds found in the work.  And, even if you get a PQR stating you can do it, most engineers will not accept the process to do their work anyway. 

Same on heavy equipment, I will use GMAW-S on some light materials on a machine, but when doing major repairs on main members I use either SMAW, GMAW spray, or my first choice- FCAW-G.  I'm not going to use GTAW on heavy equipment for MOST of the work I need to do.  Too slow and usually not hot enough. 

All processes are good for a particular application where you need to stay away from some of the others.  Then, on the next job, the ones you stayed away from are the ones you need while the others are the ones you avoid.

None of them is avoided all the time.  But they all have their place.  That's why my shop has SMAW, GTAW, FCAW (both Self and Gas shielded), and GMAW (both short and spray). 

That's also why I have Air Arc Cutting, Oxy-Fuel cutting, and Plasma Cutting.  Not to mention a bandsaw, chopsaw, cut off grinders, and an Ironworker/shear. 

No one machine or process can do everything. 

Specific for your application, without a lot more info, my first inclination would not have been to use the GTAW to weld it.  And that would be a first 'Red Flag' to me when it broke.  That doesn't mean it fit my term of being avoided like the plague because I wasn't there to know ALL the parameters of the application. 

One more, "perform as well when the weld sizes are the same?"  Yes and no.  It will depend upon rather heat input is a factor.  I can weld it faster with SMAW which in most cases is also going to contribute to keeping the heat input down.  And that can, at times, be a consideration.   More productive: most of the time, yes.

Have a Great Day,  Brent

Boon,

Your example is interesting too. I'm not a desigher but I seldom saw a design with a pad between a lug and the large pipe. Why not a lug directly on the pipe? I hope it would not be the circumstance that a designer designed a XXXX and when the XXXX hit the fan blame was put on weld contour made by the welders. In this case I would defend for the welder..."Hey, you designed a XXXX. Not me made that stinky"

If you are working on B31.1 (2010) there are design requirements Part 5, Charpter II often overlooked by shop floor people:

121 DESIGN OF PIPE SUPPORTING ELEMENTS