i read the application guide tutorial about expansion joints but about pressure balanced EJ there is only ther example of TEE and not ELBOW pressure balanced EJ.

how will change modeling with elbow instead of tee?

in tutorial the node 90 is tee intersection while i need elbow intersection: i think to model an elbow (nodes 85-->90-->120) and then attach to it a pipe (like a dummy leg) from node 90 to 95 going to the second bellow.

...........120
............|
............|
............/
A--85--90--95--B

(ignore the dots ".", needed only to align graphics)

where A,B are the two inner bellows of EJ

is this way right?

Assume you have a bend - 10-20-30.

Assume that you have a trunnion on that bend - 29-1010.

We insert expansion joints into 10-20 and 20-30, such that EJ1 is 10-15. Thus 10-20 is length of bend plus length of EJ1. 20-30 is immaterial.

We now break the trunnion so we have 29-1005-1010. 1005-1010 is EJ2 and its associated length.

Now we put in a CNODE restraint at 10 and 1010. Let's make them anchors for now, and let's say they connect to 9 and 1011, respectively.

And now we add our tie rod. That's a rigid element from 9 to 1011, which is also the length of 10-1010.

Now, we may have to change those two anchors to CNODE x/y/z/rx/ry/rz, minus one or more of these. If it allows compression but not tension (sometimes common) then one end might be a +x and the other a -x. If they allow rotation, then you might exclude rx/ry/rz. Etc. Consult your expansion joint detail accordingly (or plan to spec out the joint accordingly).

If desired/needed, you might consider replacing your one rigid element that represents x tie-rods with 3 rigid elements per tie-rod in your design.

Finally, don't forget to reduce bend flexibility by calling it two flanged (1 for the expansion joint, 2 for the trunnion). Also don't forget to calculate SIFs and input those, as well.

thank you.
but could you explain me better the last two points?

1)reduce bend flexibility by calling it two flanged (1 for the expansion joint, 2 for the trunnion).

-in which points have i to activate the flag for "two flanged"?

2)calculate SIFs and input those, as well.

-do you mean calculate bend sif? how? and then in which point i have to impose it?

3) in the presure balanced EJ with the TEE (not the bend we are talikng now) have i to activate sif in the tee point? welded type?

1) It's the bend dialogue. Unflanged, single flange, double flanged.

2) The Bend and Trunnion SIFs can be calculated via FEA, like NozzlePro. A SIF goes on the bend and a SIF goes onto the trunnion.

3) The SIF for the tee should be relatively straight forward like any other tee, unless you're aware of any special geometrical differences between it and a standard tee. It's possible that you can use tees instead of elbows with trunnions, but I can't guarantee it's cheaper to purchase and install this than it would be to purchase a license of NozzlePro if you don't have it.

QUESTION: an EJ is bought from a vendor so really we should check only forces/moments on its ends so to tell them to vendor. Vendor will check if trunnion will break or not (considering also the sif inside the trunnion). For us an EJ is an "object" able to absorb movements (and infact we model the two bellows),to push (so there is the innner pressure that will "open" the bellows), to resist (the bellows N/mm).

is it right?

I can't speak for all EJ manufacturers, but they "should" be providing you stiffness values in units force per unit length and unit moment per unit rotation.

They "should" also be providing you permissible forces or permissible displacements.

If they aren't, I cannot speak on their behalf and promise you that they're going to approve anything you send them.

That's between you, the person buying the equipment, the sales representative of the EJ, and the engineer responsible for the EJ to agree how to proceed.

Edit to add:

However, if your expansion joint stiffness is significantly lower than the rest of the unit, you can expect the loads on the unit itself to be fairly low - almost as though it's cantilevered and not attached to anything if their stiffness is 0.

At least, that's the intent.

How well real life reflects that intent (and how it should therefore be modeled) can be debated.