we had a 1" 1020 plate that had a 3/8" groove machined into it. Then a ring of 14ga stainless steel was welded over top to form a groove for a water jacket. After all welding the plate surface was machined to tolerance. For the record the water jacket is used to cool a seal. so the plate surface has to keep good tolerance
The joint was fused and to my knowledge no preheat was put into the plate.
fast forward to the first few times they cycle the unit and the weld cracked in roughly a 1' long area
My diagnosis:
martensite formation along the fuse weld because of the rapid cooling and autogenous weld between 304 and mild steel. to top it off like most fuse welds the surface profile was concave and I'm guessing the crack propogated near the most concave area. Finally the service condition is hot I think the straw that broke the camel's back was the temperature delta thermal expansion difference between the thick 1" plate and the stainless steel.
After the weld cracked. A crackshot welder attempted the repair with 308L in the afflicted area. However it cracked again right down the middle of the repair weld, and stopped right at the exact place the repair weld stopped. I wasn't called in until after it cracked a second time so the exact details of the repair are beyond me.
My second diagnosis:
weld was not fully ground out and the same stress concentrations exist, not to mention the root area was probably contaminated with porosity, rust or junk. Also 309 Is the proper filler metal for C.S to 304. Obviously the repair weld brittle as well.
Now I was called in to reccommend a solution. The most obvious solution in my mind is to completely grind out the water jacket ring, and reweld in a new one properly. However now that the plate has been machined to tolerance there's going to be issues with it going out of tolerance.
the alternative would be to cut out the offending section then repair the zone via brazing perhaps with a silicone bronze. Something that is much more elastic in the weld zone.
The machined plate costs a pretty penny so I'm not keen on scrapping it.
Any seasoned engineers have any other input? I''m pretty confident in my diagnosis but I'm less confident in a sound repair. My knowledge with crack propagation is that unless solved, it will continue to crack through the same spot, or you'll play a game of follow the crack. We've made numerous identical units and to my knowledge cracking in this area has never been problematic
Any seasoned engineers have any other input?
I have subscribed to this post :)
It's hard to diagnose problems like this from a distance, but some of your possible reasons sound probable. One thing you might try though is to use a nickel based filler metal such as 82, so the differences in thermal expansion rates are somewhat mitigated by the fact that the expansion rate of the nickel filler is in-between. Also, depending on the joint configuration, perform some sort of NDE to confirm that the cracks are removed completely.
good news, bad news.
Bad news
after careful inspection of the part it seams that the weld has cracked in multiple places around it's surface (I forgot to mention it was a lap weld of a flat laser cut ring onto a circular piece of plate.) with the extent of damage it makes no sense to try to attempt a repair weld and instead grind off the original ring and weld a new one in place.
Good news
after some talks with the chief engineer, no one could determine the choice for a stainless steel ring in the first place as the coolant channel is machined from steel and corrosion isn't a large factor. As such I believe future units will be made with a carbon steel ring which should cut out issues of differences in thermal expansion as well as reducing the possiblity of martensite formation in welding of 304 to C.S. It's just one of those things...
if the rings do for some unforeseen reason need to be made out of stainless I put in place a procedure to butter the weld zone first then weld as one normally would, and reccommended a preheat on the 1" plate
I'm still open to any discussion about carbon to stainless steel welding and repair welding for my everyone's benefit and for posterities' sake.
I think there may still be an issue as the plate will get hot and expand in spite of the cooling water, while the insert plate will be cooled by the water. Does this unit thermally cycle in use, or is the thermal cycle a once and done thing?
Can You make the whole thing out of Invar? :-)
I was going to say go to nickel too, but Greg beat me to it. If I have the configuration pictured right you have a great deal of restraint in the joint. Its actually quite close to many weld crack test configurations. The SS is going to draw a great deal and the CS won't. Ping!!!
Also, a fusion SS/CS dissimilar will have a high content of martensite throughout to be sure. Ping!!
No filler metal. Ping!!
this thing thermally cycles in service. This is the door plate of an oven retort. The oven itself operates at 900 C but out by the door the temperature is estimated to be somewhere around 100-200C.
after much talk we preheated the plate to 300 deg and replaced the 12 gauge stainless steel water jacket with 10 gauge carbon steel the same grade as the plate. That alone eliminated most of the metallurgical problems of dissimilar welding. Finally instead of fusion welding Filler was used and proper buildup was reached. The reworked welds passed test so it should be good to go now. If
If you want to pay for the Invar I'll be glad to fab it up.
After all this rework and delays I just found out that the product that is being made in the oven is so corrosive that there's only an expected service life of 6 months. The next generation will be made out of something exotic.
the oven itself is the most curious thing, The NDA was just lifted so the only just told me what it actually does.