Hi,
I have problems regarding modeling jacketed pipes in CII. As I found in CII samples, the jacketed part is fully anchored to the core at one end and is restrained with a rest and guide at the other end (CNODE). I beleive this is a proper modeling technique but:
As I reffer to CII seminar notes, it suggests full anchor for both sides which as I apply it I get wrong results ( i.e 80% sustained stress for a properly supported line. I need to mention as I remove the anchore at one end the stress turns to reach to 9%!!!) in sustained stress calculations.
I use CII 4.50 and faced this error couple of times. I think this happens due to adding too much details to the model as I think having two full anchors specially in short runs (imagine a 3x1.5 meters expansion loop using crosses not bends) is too much expectation from CII to calculate.
I beleive these strange results are due to excessive system rigidity which is in its turn because of the both sides anchores I use between jacket and core.
I'm thinking of having a full anchor at one side and a semianchor on the other side to have sth in between of both CII sample and CII seminar notes.
Let me know your thoughts

One of the typical problems with jacketed piping systems is that the jacket and core consist of different materials with different expansion coefficients. Therefore if there is any tendency to restraint (axial) thermal growth, you incur hugh loads.

The jacketed example shipped with the software is just that, a modeling example showing you the various alternatives - this is not a real job.

In a jacketed system you usually have all six degrees of freedome restrained (between the jacket and the core) at the terminal points - equipment nozzles. You need to insure that the system has adequate flexibility between these terminal points so that you don't restraint the (axial) thermal growth.


No it's not too much to calculate, you're trying to restrain thermal growth - you can't do that. You will see these large loads in the system - well until the pipe buckles anyway.

Be careful with this "semi-anchor" idea. Can this really be built? Don't just push the numbers around in CAESAR II to get the answer you want, that doesn't reflect what will happen in the field.

The layout you describe - core and jacket anchored to each other at both ends - presents high load and axial stress where there is differential (thermal) strain between core and jacket. I would not expect high stresses for the sustained case.
Either way, there's no magic here. Treat the core and jacket as simple beams and use f=kx to balance the axial forces between them.
(By the way B31.3 does not include the axial component in standard expansion stress calculations. But there is a statement in 319.2.3(c) that says you may want to consider this. And CAESAR II has a config switch to include these axial stress components.)

Hi,
Thanks for reply,
As I mentioned in my first post, I'm having problem with SUS stress not EXP.
Moreover, as far as I know there is no thermal load in SUS case. I also removeed pressures from system. Core and jacket materials are same as well.
I turn the CNODE restraints in one end from full ANC to semi ANC and 84% sustained stress disapears (Reduces to 6%)! This change shouldn't have this much effect in sustained case as I've supported the system to handle the sustained loads propely.
I think CII has limitations in applying such restraints. Or, in other words, there should be some considerations in jacketed pipe modeling using CII.
Please advise
Regards

Increasing stiffness should reduce force-based stress. You claim the opposite occurs. I can continue to offer guesses but the quickest correct answer will come from a review of your actual model and results. Would you care to send your file (the "dot C 2" file) and your CAESAR.CFG to techsupport@COADE.com?

Dear Dave,
I totally agree with you regarding this contradictory claim I'm issuing. Actually that's why I'm reflecting it here.
As I've modeled this file in the company, I can send the nutral file to the address you mentiond on Saturday evening (soonest)as the company will be closed tommorow.

I've sent the required files as you suggested.
Please let me know your idea.
Best

It looks to me as if the anchor in question is controlling the rotation about X. I built a model of this one pipe (5620-5640) using the displacements and rotations (for both ends) from the "anchor" and "no anchor" models and I can pretty much match the sustained stress ratios you mention (about 6% and 80%, respectively).
I believe the +Y restraint with a 1.5 mm gap between core and jacket at 5640 plays a role - the support is active with no anchor and inactive with anchor.
I guess the next question is can that connection at 5620 act like a full relative anchor?

I think the reason the 1.5 mm gap changes its role is due to the core deflection. As you remove rotations in ancored end the core can deflect easier ao it can deflect more than 1.5 mm touching the jacket and making the 1.5 mm support active.
The question you are mentioning is exactly what made me post this topic.
I would say it isn't a good modeling to have full anchors at both ends.
There are some cases (like what we have here) that CII can't calculate such a model as it's based on beam theory.
Please let me know your thoughts.

Arman

Is it good modeling to have anchors at both ends?
If you have the capability to control all 6 DOF on either end, then, yes, that would be a good model. My guess is you cannot ensure that the core and jacket will have NO relative deflection and rotation.
Your model had square corners without bends and those "relative" anchors at the corners. How accurate is that?
Yes, CAESAR II is based on beam theory. This works great at a system level. If you want to get a detailed, local view, then you'll probably be working with a typical finite element method. But, even then, you must be able to model the real world more accurately to get meaningful results.
I believe if you based your FEA model using the same level of detail you gave to CAESAR II, the results would not be that different.

I also beleive it wouldn't be a good modeling to fix all DOFs just because we have the core and the jacket welded with a fillet weld. I think they can have relative rotations. But it's the modeling method sugested in CII seminar notes and many engineers use it.
The reason I didn't use bends in the model is I don't have bends but crosses. I think this would be a good model for a cross fitting.

I think it is not a correct result as it happens only at the mentioned node and even worth in sutained case. Why it's much less at the other side of the loop?

If I use FEA method to calculate stresses, the results shouldn't be same as CII as the FEA model can calculate flexibilities and small relative rotations. Of course in FEA I don't fix the ends via full anchors but just a fillet weld. Therfore relative flexibilities can be calculated.

Afterall, as I mentioned in my erlier posts, in CII seminar notes it is suggested to have full anchors between core and jacket which can lead to wrong unreal results. That would be great if you could propose an alternative foe a jacketed pipe model in CII.

What would YOU suggest?

I don't know,
I was expecting more prompt response from YOUR side actually.

A bad worksman blames his tools comes to mind here. If you believe that 'in the real world' the pipe would twist or deflect more, then have you tried using thinner schedule pipe to see what the result is?
CAESAR II will only give good results with good input. Bad input - useless results.
I agree with Dave that FEA would probably give similar results here. Yes FEA is a more detailed analysis but you need more detailed input to give a more definitive answer and CAESAR II is capable.
Personally I dislike models with gaps in. Can that be reproduced on site? Are you going to check the gap is what it should be in person? By all means use it if it is a better representation, but why not set it to 1.1mm or 0.9mm etc so that it is still active, otherwise you may as well remove the guide from the system.
That's all I can say without knowing more about the problem.

Dear Perseus,
As you are mentioning in the end of your post I think you don't have enough data/insight about the problem discussed here. If you are interested, I can send you more details.
It's think we won't get much via phylosophical discussions as I'm looking for detailed prompt response. However as you took it that way I need to say a reliable model should have predictable behaviour. I don't see the model I sent to Dave a good model for jacketed pipe and I'm trying to reach more precise model (applicable in a beam theory based software like CII) via discussion here.

Any other suggestions Dave?

CAESAR II restraints (with CNodes) provide you the ablilty to define the stifffness relationship between any two beam elements - here a core and jacket. I'm working with concepts here; you've got the actual connection. It's up to you to enter that stiffness. Typically, where thermal strain is the issue, stiffer is more conservative.