API 650 - Structural to Shell Connections: Pinned vs. Fixed

[THef]

Good Morning Everyone,

I'm checking some pipe supports on an API 650 tank and they are all knee braces (some supporting two pipes so moments can be present). Since the shell plate deflects under loading, I'm inclined to consider it a pinned connection (despite the fully welded connection). However, adding structural and/or repads (to distribute the load into a larger portion of the shell to lower my shell stress) will stiffen the shell. Now, it still wouldn't be a fixed connection, however if I do check it as a fixed connection, I find the moment loading into the shell causes my shell stress to be too high with my current details at a couple locations.

So, when attaching structural to the shell, do you typically treat it as a fixed or pinned connection? I feel like I should go with some sort of semirigid connection and perhaps use a portion of the moment from the fixed connection, but I've never done that kind of design and wouldn't know where to begin.

And, just out of curiousity, if they were cantilevered supports, would you change your decision?

Let me know what you think, cheers,
THef

 

[pennpiper]

It might help you get an answer if you gave a little more information.

What is the size of these two lines?
What is the commodity?
Will they be Hydrotested?

Are these lines traveling vertically up the side of the tank? And if so, how high up do they travel?

or

Are these lines traveling horizontally around the tank? And if horizontal, how far around the tank and what is the spacing of the pipe supports?

 

[THef]

Good Morning pennpiper,

I'm more interested to find out how other people consider loading into the shell generally than in the specific cases I'm looking at (which will actually be everything from vertical overflows to circumferential fire suppresion systems).

We've only recently switched from ASD to LSD (LRFD in the US) for structural design. With the need to check multiple load cases to satisfy LSD, as well as still running an ASD check for the actual loads on the shell, it's faster in RISA than by hand. Since we're changing our methods, I thought I would reexamine our logic as well to ensure it makes sense.

I believe that it will still act as a pinned connection unless you're at an incredibly stiffened area, but wanted to know if that's the consensus or I'm out in left field.

Regards,
THef

 

[Tankman650]

THef,

I would use pinned connections if it is at all possible. What stress alowable are you using? For and API tank, S is given in API. For attachements you can have local stresses of up to 3S for local bending and 1.5S (REF: ASME VIII, DIV 1)for simple tension/compression. When using any general FEA model, use lots of elements to get a good picture of the local stresses.