Dear all,
I prepared one stress model which is very big. While running the model I am getting problem in iteration. For that reason I break the model into two parts.
After breaking the model into two parts, the end point of 1st part will be the start point of the second part model.
For more clear i would say my 1st part model starts from node 10 and ends with node 800. Second part model start with node 800 and ends with the node 2600 (which is the end of the model).
My question is while running the each model how should i end the Node 800 and for the second part how should i start the second model?
As a whole model at node 800 i have used one Restraint that is +y. so which type of restraint i will use now at node 800 after breaking it into 2 parts?
Shall i use anchor? Or else please clarify?
I hope my question is clear. Please tell me the procedure.
Thanking you all.
Please help me in sorting out my problem.

Peterson,

Have you tried solving iteration problem by removing friction for lifting supports, using Y in lieu of +Y for supports that rest & making gaps zero for some supports which are causing prob in iteration!!

If these don't help & if I'm interpretting you correctly (I hope I'm :-0)), you've made two Caesar files by breaking down the model into two...right?

Run any one of the model say you run the model from nodes 10-800 (of your example). From the displacement report, get the displacements & rotations for node 800 & impose them at node 800 in your other model for correct results.

But I would still prefer that you don't break the model (as your system is connected & not isolated from each other) so that you get accurate results.Or if you can isolate the two models by putting anchor or triaxial support at node 800, it would be better.!!

Thank you very much SJ
First I will try to resolve the iteration problem. If not then the second method.

Thank you once again.

Thank you SJ I solve the iteration problem.

Peterson,

That's good!!

Additionally, there is one more method to solve iteration problem due to friction that is by changing the "friction stiffness" under "convergence tolerances" in cfg file when none of the above methods work.However, I've sometimes observed that this may lead to change in loads on supports & nozzles (sometimes this change is too high).Use this as the last resort but be cautious when you use it!!

http://65.57.255.42/ubbthreads/ubbthread...=true#Post42617


http://65.57.255.42/ubbthreads/ubbthread...=true#Post38355

You may find the above links regarding friction stiffness

Search "decoupling". There are some good posts.

Regards,

Good day Peterson,

If you need to split large stress system into two, you need to locate this break at the most suitable boundary condition such as equipment nozzle or piping with 3-way supports (rest support + guide + limit stop). In this way, you are considering the near realistic zero displacement at that point. For support loading, need to do algebraic sum of loads taken from the result of each smaller systems. In your case, node 800 has +Y only and it means piping is free to move in X and Z. Need to take care also of the 3 rotational movements (Rx, Ry and Rz). See "decoupling".


With regards to changing friction stiffness, check first Ceasar II Help for the consequence of lowering this value. I rather check the liftoffs and gaps in every support than changing this default value, unless I understand that this will represent actual behavior of piping and supporting structure.

Regards.

thank FKX for your opinion.

regards.

SJ, when you say, "Have you tried solving iteration problem by removing friction for lifting supports, using Y in lieu of +Y for supports that rest & making gaps zero for some supports which are causing prob in iteration!!"

How do you find the supports which are causing the problem?

Thanks,
- Jonathan

Press F2 during the analysis to see what's not converging on the current iteration (red arrow below).
The blue arrow on screen below shows that the Y restraints at 262 and 264 are not satisfying friction criteria and the (+)Y support at 282 is not satisfying the liftoff/resting criteria for this iteration.
I prefer to work with liftoff/resting supports and gapped guides in resolving such convergence issues.

Could you define liftoff/resting supports? I searched for it in the CAESAR II user's guide and I could not seem to find it.

Dave:
Are the elements that are not converging the ones that have "ERR=" and then a number or are all of the elements that show not converging?

Is there material that I could read explaining what is noted in the dialog box?

The SLIDING ERR= refers to nonconverged friction terms. There are two nonlinear parameters with friction - magnitude (as N can change between iterations) and direction (as the sliding path - and applied friction load - can change between iterations). Friction also has the stick/slip nonlinearity. With all this going on, I usually try to avoid addressing these issues and focus on simpler nonlinearities that I can evaluate when viewing results. {You can also change the friction tolerances for convergence but that adds more uncertainty.)
The listed items that do not have that ERR= are basic ON/OFF situations - the restraint is active or the restraint is inactive, or, RESTING or LIFTOFF for a +Y restraint.
A pipe sitting on a beam would be modeled as a +Y restraint - the pipe is free to move up (LIFTOFF) but restrained from moving down (RESTING).

The further I worked with the errors per non-convergent case I realized that I could eleminate the friction coefficient at that given error node and repeat for every case. It is now working.

I appreciate your explanation and help in this matter. It makes more sense now.

Eliminating "mu" at those locations removes friction - not usually the best thing to do. You can search this Forum for the topic "convergence" for other recommendations. I would suggest tweaking the "Friction Stiffness" in the Configuration before removing friction from the restraints.

I understand that it removes the friction from the restraint. I have read quite a few topics on this issue.

In one reply that Dave gave he stated that he had changed the friction stiffness and CAESAR II had given an unreasonable analysis. This analysis was traced back to changing the friction stiffness.

I have eliminated the "mu" on 30 of 500 restraints. I am trying to find the best solution with the means provided.

If I were to change the stiffness of the friction I would not know where to start. Could you point me to some literature on the subject so I may try and educate myself.

Thanks,

There is nothing in the literature on this. Convergence is a balancing act, based on the behavior of the system.