Hi all
I am using tied expansion bellow in my system. Here I got some doubt regarding modelling procedure.
I fell we have to block all rotation I,e Rx, Ry & Rz . Please see attached sketch. I think we should make free only 2 direction movement . Please correct me if my understanding is wrong.

Shr,

This approach is not correct. Tied bellows can also take some torque although very small. By putting RY in your model the torque acting on the bellow will be affected by the stiffness of the rigid element,which will not give you correct values.
Best thing is to check the rotations (RY) with vendor allowed values.

Regards
Siv

To add any Ry, Rx or Rz into tied expansion joint model is wrong for me. If you dig out old original design documentation for CAESAR II (DOS times) you will see what is happening. Also if you have 5.1 latest release c2ap.pdf Expansion joint section is revised. With quick checking number of rods issue is addressed to some extend. CAESAR II is clever. If you have 2 rods it prevents automatically end rotation around axis at 90 degree to a line connecting the two rod ends. If you have 3 or 4 rods ends will not rotate if the pressure is positive, e.g. rods are on tension (Or compression). If you add rotation restraint you most probably get wrong calculation results. The trick, which is overlooked by designers is that if you use 2 rods you have to specify rod orientation for erection people and bellows manufacturer.

Torsion is different issue. As mentioned do not put restraint. Tied compensator has high torsional stiffness but adding the restraint your model is wrong. Get the stiffness from the supplier and at the same time maximum moment they allow (EJMA code has formula for it). Basic rule is that there should not be a torsion moment. Bellows fail easily under torsion and there is no way to design tied compensator with built in restraints that carry the torsion.

Thanks Jouko & Siv
I agree tied bellow is not design to take high tortional load. Here my intention is just to calculate tortional load in more conservative way not let a tied bellow to place to take more tortional load. As for example refer attached sketch. I will get more tortional load so I must be more cautious in my system. In this case angular theoritical rotation will be zero. I feel if I try to restrict tortional moment there should not be any issue.

Richard whether the modelling given in CII document is it wrong????

The way I see it - the XJ manufacturer provides maximum torsional rotation allowed for the joint. The XJ manufacturer does not tell you how much torsion load the joint can take.
By adding your RY, you will get the load but not the torsional rotation. What do you do with that number?
I would omit the RY and work with the rotation limit.
You may need to reduce the torsional moment but I believe working with the rotation will make your task easier.

No I don't think it was wrong. We had a number of suggestions to improve this section of the documentation to make things clearer and more explicit.

Thanks everybody for response.
I totally agree with Dave.Again when I am putting tied expansion bellow at very sensitive nozzle sometimes may be tortional load more important than than tortional rotation.Otherwise yes we need to know tortional rotation to check bellow limit.
So can we change tied bellow modelling technique case by case basis.What's your opinion?

I had another thought about this last night.
Yes, I still believe it's more sensible to work with torsional rotation BUT you should have a good idea of what the torsional stiffness of the joint really is. I look at the Senior Flexonics Pathway catalog and I see no torsional stiffness in the tables but if you use their XJ modeler in CAESAR II, you will see that they give us a value.
You may have to dig around for this number.

EJMA code has formulas to calculate theoretical bellows shear stress and angle of twist, which you can then use to calculate the spring rate. Formulas are in C-4.2.4 of the 8th edition. Included into the formulas is the torsional moment. Pathway is using these formulas, my software is also. If you are not a designer/manufacturer the formulas do not help a lot as there are parameters that are not published in most of the documents. Pathway is a well known manufacturer and you can trust their values but they are valid for their designs only. Other suppliers' values are of similar magnitude.

EJMA code limits the shear stress to 25% of the allowable! Bellows do not like torsion.

When I have CAESAR II model with compensator. I take out the moment and use that as an entry value to EJMA formulas to check that the element is not overstressed. A bit of iteration is required as the spring rate must be calculated first for CAESAR II input.

look at the expansion joint rating module of C... find it under the name analysis....

I recently had a discussion that a lateral expension joint can also handle a little bending (reducing the max lateral movement of course). This seemed to be a very strange case to model cause it can happen, but most probably will not happen (without breaking the bellow).
Always check what a bellow can and can't do. Then check whether or not the movement is already blocked with a 1e12 spring value in the bellow settings and if not, add a restraint.

Senior Flexonics Pathways provides Torsional spring rate and allowable torsion angle in the back of their catalog; page 68+/- depending on the catalog date.

If the bellows designer knows that lateral compensator has also bending movement it is quite easy to design. Bellows must have only 2 rods and then the bending is possible. To model it in CAESAR II is easy. Model as normal without Rx, Ry or Rz and make sure the rods are in correct plane. CAESAR II handles the rest.

If values are available from a supplier it is a good idea to use more conservative values. As a general rule:
- it is easier to design long than short compensator. Allow longer lenght
- it is easier to design bellows with higher spring rate than lower. Use higher rates
- compensators can be heavy. Allow higher mass
- lower the maximum torsion moment

Other points
- never include unnecessary safety margin to required movements
- if the movement is due to thermal expansion it is a very rare case that you need to specify equal compression and expansion or equal lateral movement in 2 directions. It is perfectly OK to specify lateral movement 60 mm East, 10 mm West.
- never overspecify required fatigue cycles. Remember that EJMA and B31.3 use different theory for fatigue. Specified number of cycles to B31.3 should be lower than if bellows is to be specified to EJMA.

Each bellows manufacturer have their own tooling and method of manufacturing. Using values from one manufacturer may exclude an other supplier.

A bit out of topic, sorry for that.

“If the bellows designer knows that lateral compensator has also bending movement it is quite easy to design. Bellows must have only 2 rods and then the bending is possible”
This option sometimes I use to reduce bending moment in one direction though I was not very much confident on the same . Instead of blocking 2 directional bending moment and axial displacement as mention in caesar II tied bellow modelling I use to block one direction bending moment & axial displacement by tie rod & use to specify tied bellow with two tie rod unit in bellow specification & mark direction of tie rod position in piping general arrangement drawing & piping isometric. Is this approach reasonable when I have small amount of bending but still not ok for equipment nozzle load to block that small amount of bending?
Do I need to inform bellow vendor bending angle my system is getting for their design?
Again thanks everybody for giving your valuable comments on the subject.

What I saw it seems that CAESAR II documentation is amanded in to some extend. Very old documents indicate that you do not use any Rx, Ry or Rz. My calculations, which I was able to check on site confirmed this. If you have 2 rod arrangement I do not see that the calculation is out if you use one rotational restraint provided it is in plane where the rods are. Just unnecessary work.

You are 100% correct in showing the rod orientation. With two rods and rotation it is important.

You have to tell the supplier about the bending movement. It has to be considered when the bellows design is done. Relatively small angle can have substantial impact on bellows design. An other issue designer has to know if centre pipe (nipple) expansion is included or not into the bellows movement values. On a universal bellows where rods are under stress always not critical. Designer can normally figure it out but where the rods are limit rods he will not know.

Thanks Jouko
Your advice & suggestion is really helpful to us.