FEA on pipe flange below MDMT

[PaperEngr]

We have some existing A105N, B16.5 flanges that have not been impact tested. New upset conditions would be -48C at nearly atmospheric pressure for 15-30 minutes. Most of the time it will be at ambient. We have done pipe stress analysis and checked flange leakage, both are fine based on B31.3. Client now wants to do an FEA.

I can barely spell FEA, so my questions are many: Is there anything to be gained from the FEA? Obviously the stresses are going to be high around the bolts - I'm sure this will only raise more questions. Can an FEA calculate the temperature gradient from inside the pipe to the outside edge of the flange? Can an FEA take into account the material going into a brittle transition zone (MDMT is -29C per Code)?

Thanks for your time.

 

[TGS4]

An FEA will provide you with no additional pertinent information.

And an FEA on a bolted flange joint will provide you with even less and will likely muddy the waters.

What you have done is sufficient.

If you called me asking me to perform said analysis (something that I do on a frequent basis) I would turn down the consulting opportunity with exact same advice.

 

[PVPguy]

Respectfully disagree with TGS4 on this one. FEA will get you a few things:
1) You can calculate the thermal transient response of the flange to the upset condition. This will give you the metal temperature of the flange during the upset. Depending on the size/class of the flange it will take time to respond to the thermal upset and so the metal temperature may well be above your -29 MDMT for a substantial portion of the flange, allowing you to deem it acceptable.
2) It will give you the stresses in the flange and while they are unlikely to allow any temperature reduction (based on the stress caused by assembly bolt load) it is a useful input (potentially for proving why the upset is actually of concern for brittle fracture).
3) It will allow you to determine if the joint is at risk of leakage during the upset, if that is of concern to you.

Of course, all of the above is only relevant if you know what you are doing with the FEA. There are also non-FEA methods that can be used (particularly as a first pass). Look for WRC bulletin 510, for example.

 

[PaperEngr]

Thanks to you both for your input. We'll have to wait and see what the Client chooses to do.