I am analysing a line which a fairly big model with 150 over nodes and 50 over support.

During anlysis I have to go thro this agony of removing all the non-linearities viz, the friction factor and guide gaps which I had given with so much of care.Normally removing the friction or gap from the first of the "non converged last iteration" nodes listed in the INCORE SOLVER resolves the matter.But in this system of mine it was a literal wild goose chase , I ended up removing the friction/gaps of more than 20 restraint points.I would like to know if any body knows to work-around this problem in a more scientific way than me . I am sure there must be one bcos the INCORE SOLVER menu ,in general and those input fields like "Normal load Variation", "Sliding angle variation","Sliding Force multiplier" and the button "default values" and terms like "old state close neg " , "new state closed neg sliding err = 0.568" etc must be meaning something to more enlightened souls out there than "the most common use (misuse) of the INCORE SOLVER , to show that ur working hard even when ur not at ur seat " .

S.C.Rao

Tessag Edeleanu Asia Pte Ltd
http://www.edeleanu.com.sg/

My first recommendation is to review the write-up in the CAESAR II Technical Reference Manual on the Friction Solution. This will give you an understanding of how the solution works, and what the terms mean.

When approaching a friction problem I suggest two models, one with friction, and one without. The model without friction can be used to verify everything falls out the way you think it should, and you have no gross modeling mistakes. The second model is where you add friction. Two things to note here: (1) friction can only be used to hurt you, you can never count on it to help, (2) be careful of your "mu" value - it will likely vary over the life of your piping system.

Now, how do you go about solving a convergence problem?

<font color="#0000ff">First:</font>
Use the [F2] option to single step through the restraints to deterine what isn't converging. I admit the resulting display is somewhat cryptic, but ...

- The three numeric values (such as 0., 1., 0.) are the direction cosines of the restraint. This is an example of a "Y" direction restraint.
- Old state "open", new state "closed", or visa versa indicates the pipe is bouncing, such as on a "+Y" or in a gap.

- "err = xxx" is the percentage change in the normal force on a friction restraint.

<font color="#0000ff">Second:</font>
Identify the nodes with problems. Specifically target gaps and friction. Extremely small gaps can't be built and should be either removed or enlarged. Friction issues take a little more effort...

<font color="#0000ff">Third:</font>
Rather than altering one restraint at a time, it is often better to adjust the friction parameters for the entire job. This is done via the "Configuration Module", on the "Computation Controls" tab.

Friction Stiffness: The default value is 1,000,000 lb/in. Reduce this by half and try the run again. Continue reducing this by half, but I wouldn't go much below 25,000 lb/in, it makes the system too sloppy.

Normal Force Variation: Increase this in steps of 5%. Remember, the larger this value, the more error in your solution you allow.

Angle Variation: Increase this in steps of 5 deg. Remember, the larger this value, the more error in your solution you allow.

<font color="#0000ff">Fourth:</font>
Remember (from the documentation) that there are multiple equilibrium solutions to a friction problem. Review your results and ensure they make sense. Are they in-line with your "no friction" run? What happens if you change "mu"?

Friction problems are not trivial and require a great deal of thought. I hope this information helps.