Crack and gas leak from a SS 304H to SS304 weld with filler ER-308H

[QCJT]

Everything seems allright with the above combination but could it be one of the following:

- It was a fillet joint connecting a pig tail elbow (3mm thick) to a lifting lug (8mm thick) with a throat thickness of 3mm - too small?
- The system is a set of headers (4" and 6") hanging on spring supports connected with the above mentioned lifting lug - spring loads weren't considered for the fillet weld size?

Any other thoughts?

 

[metengr]

Could have been a solidification crack caused by little to no ferrite in the weld deposit.

 

[stanweld]

Was crack in the weld metal? If so, agree with metengr.

 

[jtseng123]

Have you checked the weld strength ? what is the design temp ? Creep rupture can be a problem.
We now use 16-8-2 instead of 308H for load bearing components at high temp service.

 

[QCJT]

jtseng123 - The operating temp at this part is around 520 deg.C. How shall I check the weld strength? The PQR was of course tensile tested, if that's what you mean. But since this is a fillet weld it was qualified by a butt joint and we didn't do any specific testing for this kind of joint - but I think we should do it now because the type of joint and the loads are very different here. Is this what you meant?
What is 16-8-2 - could you give me more details on this filler please?

Metengr, stanweld wouldn't solidification cracking be apparent almost immediately? This thing was in service for 3-4 months.. I will only get to know this after we conduct LPT when the furnace has cooled down (we had to shut it down due to this problem). Then if we find the crack is in the weld metal we might try to do a PMI of the weld pool to get the ferrite content.

Isn't anybody considering that the fillet weld throat was too small? They used a standard formula for deriving the fillet weld throat (a=0.7xS, S=thickness of thinnest member), which gave a result of 3mm throat. This looks for me too small and it was not a full pen weld even. I mean these joints are basically supporting a system of headers, which seems way too heavy for this kind of weld. The thing is that our engineering didn't even see the contractor's shop drawings and thus didn't cross check their calculations...

 

[jtseng123]

520 C is not high. I am dealing with 780C all the time. The use of 308H could be fine in your case. Google "16-8-2 weld metal" you will know what it is, and it has comparison with 308H or 304H base metal regarding the creep rupture.

I am a mechanical engineer, so I can only talk about stress and design. I have all kinds of lug design through the entire petrochemical plant for years. I always ask MT or PT root and final passes for fillet welds and well calculate the stress on support lugs, no exception, and never has any lug failed. You keep mentioning the weld is too small, so the questions to you still remain the same:
What is the weld stress? ( it is somehow complicate calculation if it involved bending, but not difficult)
What NDE has been done ?
You can investigate the ferrite as mentioned by metenger. But you shall start the easy ones the two 'Whats' I mention to exclude basic mechanical problem.

 

[metengr]

Your original post did not mention this was in-service. Second, weld solidifcation cracking can be subsurface and propagate to the outer surface in service. The fillet weld size does seem rather small but where exactly is the crack or defect? You did not mention the location.

 

[QCJT]

It became now evident that the crack isn't in the weld pool it's right at the border line between the weld pool and base metal, i.e. the crack separated the weld totally from the line.
One issue to be kept in mind for us is that the cracks are only on those pipes where we had a by-pass line welded onto same manifold, very near this lug/elbow joint and this by-pass line remained cold during operation, while the rest of the line had process temp of 520C, i.e. no expansion with the rest of the manifold, which we think now might have been the reason of the problem and perhaps will get rid of this by-pass