I have seen this same (or very similar) image and question in the past. Since this seems to be a sample (maybe for qualification)  and not a production part, some investigation is required. I have seen cases of this on Welder Qualifications for test welds however without the thickness being reduced below the base metal thickness.

D17.1 may not have the "support" you need for rejection however if you don't wanna put your name on it and can explain why, just go for it with the understanding that you are of course accountable for your actions. The other option is to place a hold on the item and request guidance from the engineering authority based on the fact that this condition has reduced the effective thickness of the base metal/weld metal below the as supplied thickness.

As an inspector, for all I know, the design "minimum" thickness is 75% of the nominal thickness. Thus the joint MAY be OK however I would request guidance from the engineering authority 1st.

Codes do not always cover all the bases but if all you have is the code and you wanna stick with the code, stick with it.

I probably wasn't much help. Some of these things are best answered internally. Hate to have to back up a questionable decision with "...well the guy I talked to on the internet said .........."

Have a great day

Gerald

The call-out may have been for a fillet.

What you have achieved/shown in your excellent image is a T-Weld with melt-thru (Fig. I-2) That has a noticeable/measurable reduction in t

Since your joint is actually CJP,  I think you can make an argument to inspect it as such and reject for underfill on the "other" side and undercut on the arrow side (unacceptable profile via Fig 6.1 [C].

Fillet or CJP T-Weld is not a stopper here
Table 7.1 uses the terms  "Underfull and/or Concavity"  This is what you can rest on.  They are two different things, and the code is recognizing this.   The back of that weld is concave and measurable.  You have a formula to judge via weld class in the table.

I don't have a coy of D17.1 with me, but isn't there a clause in there somewhere for base metal damage (nick, gouges, etc.) beyond 10% material thickness to require rejection / repair?

Tim

I would classify this condition as underfill, because, although Table 7.1 in D17.1 says "Face or Root Underfill (Groove welds)", Clause 3 Terms and Definitions defines underfill as " A condition in which the weld face or root surface extends below the adjacent surface of the base metal."  The definition of underfill in D17.1 does not specify the condition applying only to groove welds, so I believe the argument could be made in the case of the supplied picture the root surface of the weld extends below the adjacent surface of the base material, and could be evaluated using the underfill acceptance criteria in Table 7.1

metllab-

In these cases ---> http://imgur.com/a/QJu4h    I rejected these.  I ran into the same dilemma.  Now if this is to D17.1 CLASS A I would reject based on undercut since its .002 based on full length of weld.  Also as was said before you can make the argument that you are checking a fillet weld which is clearly no longer the joint. It is now a full CJP.  I would send the customer the photo along with your discrepancy and see how they handle it.  You cant judge a fillet when there is not one.  Also I believe the minimum unspecified weld size is 1.5T.  that weld does not look like its more than 1.5T the vertical material.  my 2 cents

J Maxwell

I would have a difficult time rejecting a welder qualification test for a FILLET welded sample based upon the criteria in the code only for the condition shown if the undercut on the welded side of the joint did not exceed .07t. The condition on the other side is not undercut as defined by any sources I am aware of.

I think this very condition (which can be difficult to control with some production settings) is why 17.1 allows for some IP at the root for thinner materials with fillet when performing qualification tests.  But that is just a guess.

So suggesting to your client that "cooler parameters" could prevent/minimize this type of condition could be something to minimize this is the future.

Mil-STD-1595 used to address this specifically :(