Can you provide more information as to where in the fabricated T the cracking occurs? Is it where the nozzle is located or in the main pipe run butt welds?
OK guys picture a column that is 5/16" thick with ribs attached to it that are 7/16" thick. Where they connect there are 5/16" full penetration welds. The failure is occurring on the HAZ of the 7/16" ribs. I'm wondering if the full penetration welds are necessary, maybe just a fillet weld would suffice. The other end of the ribs are attached to a smaller diameter column where a shaft runs through it.
There are no reinforcing fillet welds and the cracking takes place at the toe of the weld of the rib. Under normal circumstances I would agree and say it's fatigue cracking, but after column is in place and exposed to vibrations it either cracks relatively early or it will last for many many years.
After fabrcation do you perform a Liquid Penetrant inspection of the completed welds? If not, you should do so.
Second comment - low cycle fatigue could be the cause of cracking. If the severity of stress or strain amplitude is below a threshold, there will be no source for fatigue crack initiation or propagation.
m3bruof,
There are a number of welding paramters that can affect welding residual stress. This would explain why some fail in low cycle fatigue and some do not. Reinforcing fillets are advised. If reinforcing fillets cannot be employed, you may wish to peen to improve fatigue life and lower the residual tension weld stress at the crack origin site.
This case sounds like it has all the characteristics of a classic small nozzle fracture. If it is indeed a case of fatigue, then there are several possible angles of attack:
1) Reduce the amplitude of the vibration, by damping it at its source and/or somewhere along its path to the fracture location.
2) Add gussets (if feasible) that act normally to the plane of vibration.
3) Reduce the geometric stress raiser by ensuring no undercut and then grinding a radius at the weld toe. Peening would be a further measure.
Note that adding more weld will not help; this is a base metal failure. The weld merely provides a stress raiser conveniently coinciding with the point of maximum stress caused by bending. In fact, it can exacerbate the situation by increasing the tensile residual stress.
