316/316L Flange Distortion After Welding

[monty011]

Here is one from a previously closed thread. We are having issues with flanges distorting (convex).

The joint is comprised of a plate flange with weld detail that would typically be used for a slip-on. The flange is 1.00" thick, and the pipe pup is 3/8" thick. The welds are called out as 7/16" fillets. The weld sequence has been done both ways (inside weld first, and inside weld last) with no change. The flange is still distorting.

Normally, i would think that the OD weld would be the culprit (pulling the flange downward upon cooling). But, since this is a plate flange with no stiff hub, could it be the ID weld that is actually distorting by heat input? The weld is located off the neutral axis of the plate (centerplane) in this case.

If the weld details can't really be altered, i'm not sure how to limit distortion without bringing in a strong back or other method of fixturing. Anybody have any thoughts/experience on this?

 

[berkshire]

monty011,
How flat is the flange before you start welding?
Typically if the flange is doming you have excess material in the center requiring the outside of the flange to be stretched to regain flatness.
Also how is the flange fabricated , is it profile cut from plate, Have you checked for residual stresses from the cutting operation, typically the profile cutter will cut the hole first , then the outside. This may build up stresses in outside of the part that are relieved by your welding operation allowing the part to dome. If you pre heat the flange and allow it to cool without welding does it stay flat, or does it dome, or potato chip ?
The other question, does post weld heat treatment make the problem better or worse?
B.E.

You are judged not by what you know, but by what you can do.

 

In my experience, donut shaped welds are the worst case for residual stress and potential for out of plane distortion. And you are working on a class of material that is the most prone to distortion.
This looks not so much like a case of angular distortion (as you would expect in a linear joint between plates) as it is the weld pulling radially inward. This tends to cause flat shapes like your flange to dish. Which weld you make first won't change that significantly.
I concur with berk's suggestion that there may have been residual stress locked up in the flange plate from thermal cutting. This could be mitigated (slightly) by preheating before thermal cutting, but far better would be water jet cutting.
When all else fails, make the flange thicker and machine it after welding (if that's feasible, but adds cost). Almost as a rule you do not want to stress relieve austenitic stainless steels.

Your only other weapons are: welding process (some are more prone to distortion than others); bead size; sequence; and redesign.
Avoid stick welding and go with a semiautomatic process, GMAW or FCAW.
Contrary to popular opinion, many small beads make distortion worse - a reason not to use SMAW. (I call this "trying to sneak up on a weld".) Bead size should be as large as practicable, and keep stop/starts to a minimum. Avoid weaving. The welding is best done using a rotator.
Sequence - these will be multipass welds, so balance the welding between the two fillets.
By all means do NOT overweld.
Can the fillet size be made smaller?
Preheat can modestly mitigate distortion, but with 316 you have limited headroom.


"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[racookpe1978]

Yes, weld up the assembly, then machine the mating surface flat and true.

This may require an added thickness on the initial flange to allow a machining margin, but your not buying an off-the-catalog pre-fabricated flange anyway, so the extra material cost will be minor compared to losing the whole flange.

The final weld will "pull towards" the heat, so alternate the inside and outside welds. The 7/16 fillet is very, very likely to be oversized, and that near-1/2 fillet is very, very hard to avoid bowing if one side at a time is welded. tack it on both sides, then do welds on one side of only 1-2 inches long alternating sides with each intermittent bead until 1/4 fillets are made-up all around both inside and outside. Only then finish the welds, and try to alternate at 6-8 inch lengths.

 

[monty011]

Good info, guys! Just to add some information on the flange. It is water jet cut from plate and machined down to size. Originally, they purchased a net sized flange, then used FCAW on a rotating positioner. The flange cupped, and I sent the both flange and reducer it was welded to out for solution anneal. The annealing brought it back to about 50% - 60% of the way home, so there was a fair amount of elastic stress built up and unfortunately, a fair amount of plastic.

I had them cut the flange off and try again with adjusting some weld parameters, preheating a bit and having them mind the heat input. The new flange was ordered quite a bit thicker in case we couldn't get rid of the distortion, and sure enough, here we are.

Looks like we are going to go the same route again since the thicker flange has cupped in the same manner. So, we will solution anneal again and machine the faces down to spec.

 

Post-weld solution annealing will disqualify your WPS/PQR (unless it was qualified that way, a very uncommon event).
Post-weld heat treatment, including solution annealing, will never reverse all the distortion caused by welding (it is another popular assumption that PWHT can cover all the sins of welding).
If it remains too difficult, consider changing to a weld-neck flange. More costly I know, but in the end less cost overall and certainly less grief. Purchasing agents will object, but sometimes you need to spend your way to cost savings.


"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[monty011]

In our case, it's a pretty common occurrence to solution anneal post-weld for NACE compliance. So we have to be covered there.

 

[TimSchrader2]

Hello

We Weld 3/4 pl or 1/2"Pl to 1.5"Pl in a T fashion and the welds cause similiar distortion. We pre bend the Horz pl (OR FLG in your case) to minimze bend distortion. There is a way to calculate this, but is very aprox. A have a tech sheet around here in a file if needed.
Machining is req if a real flat surface is req. As mentioned.

 

[monty011]

Thanks Tim, I believe I have those approximations in an AISC document somewhere. I guess there's just no getting around distortion with aus. stainless. I think there are some changes that we will have to adopt on the process side, similar to how you guys are pre-forming prior to weld.