View Full Version : TIG'd T304 Failure, Embrittlement, Cracking, Rust
06-17-2009, 01:19 PM
I have got a turbo exhaust downpipe that I purchased a while back and it has cracked around one of the weld beads near the turbo outlet. It was supposed to be T-304 stainless, but it has a brassy tint to it (pretty normal) with a number of patches of what appear to be carbon black rust. One of the O2 sensor bungs appears to have sprouted some of this and the entire weld bead is a black oxidized/rusted entirely.
I'm not sure if I just got it too hot with retarded timing (ironically for safety) or if it's a material issue or perhaps a lack of back gassing on the tig welds at the el joints causing embrittlement. Perhaps the O2 sensor bung is of dissimilar material?
Any thoughts? I posted this over in the automotive area but I haven't heard back yet. I'm including some photos as well here of the sensor bung and the crack in the pipe.
Thanks for your time,
Stick a magnet on the bung, that will give you your answer there.
Given the size of those welds, the fact that several dimples/craters are visible in the bead, and that there appears to be undercut and underfill both, I'd bet that a crack started either at a crater or weld toe and then just propagated right on down the line for the bulk of it.
06-17-2009, 02:51 PM
It's tough to say for certain what happened from the photos you show.(not your fault, most folks don't have a microscope)
But there are a few things I see that concern me. Looks like you've got fractures running down the centerline of the weld bead and along the edge, perhaps in the HAZ. Can't tell where it started from, but failures in the weld bead are abnormal.
I do not like the weld bead appearance. Looks like the weld is underfilled/undercut. It also looks like the weld was formed by tacking around the joint. You can see a 'nipple' in the center of each ripple of the bead. Each one of those spots is a potential pinhole through the weld. Even if no holes penetrate completely through, there's probably a small cavity on the inside, and each cavity is a thin spot in the weld. The weld shown near the O2 sensor bung is identical. The other weld failed first, perhaps because it's exposed to higher temps or more stress from engine vibration?
The soot combined with high temperatures, is a perfect recipe for chromium carbide precipitation in the steel. Mix in a little water from combustion by-product, and your welds will corrode away from the inside out 1/10th the normal time for SS to corrode. Normally this corrosion proceeds slowly, but give it a crack to start in and it can take a fraction of the time.
What was the lifespan of the system? How many hours of use? How many temperature cycles did it see before failure?
So, with the information available, I think the likely root cause is underfill in the weldment. This, alone, or combined with a likely pinhole in the multitude of small tackwelds, started a crack in the weld. Thermal fatigue and/or fatigue from engine vibration caused the crack to grow. It ultimately failed by cracking around the weld; the crack leaping from pinhole to pinhole or following the path of least resistance along the thin edge of the weld bead.
The fix is to use a better welding technique with proper reinforcement. Don't weld by making a series of tack welds around the weld joint.
Make 2 passes if necessary to build up the weld and get good reinforcment on the face and root.
Ensure the welder uses a low carbon filler metal; SS308L or SS316L would work nicely. This will resist carbide precipitation and accelerated corrosion of the weld and/or HAZ.
Make sure to protect the backside of the weld by Ar purge or some other method that blocks Oxygen from the weld.
06-17-2009, 05:17 PM
I would go with the common theme here; the weld did not have enough filler, if any was used. The weld should be for sure backpurged for this application. Depending on stress, load, cycling, harmonic vibration, the tube may have issues trying to stay together. Basically, A Dab Will Do hit it pretty square on the head.
vBulletin® v3.8.7, Copyright ©2000-2013, vBulletin Solutions, Inc.