I'm sure this topic has been done to death, but as a relatively new user, I've got a bit of catching up to do.
I'm working with a large model produced by a subcontractor, which has several non-linear restraints. When I try to make a small geometry modification and re-run the model, it does not converge for certain load cases.
I've tried adjusting the friction stiffness in the config setup window, but this doesn't seem to help, even when reduced to very low values. With some load cases (but not all), I can get the model to converge by increasing the Friction Normal Force Variation up to 100% (although I note the help file admonishment that this value should not be adjusted from 15%).
I can obviously get the model to run by reducing friction at the offending supports to 0, but this does not seem to me to be an ideal solution.
Does anyone have any comments or experience with this problem?

Adding friction to a job makes the job highly non-linear. There is a discussion on the solution to the "friction problem" in the CAESAR II Technical Reference Manual, P6-16. In general, the steps to achieving convergence in a "friction problem" are:

1) Determine what load case isn't converging and make sure it is realistic. By this, I mean don't setup a load case with only a single load component such as WIN1. Why? Well, wind will never act alone, and for code compliance, you should really obtain the effects of wind by subtracting two operating cases, one with wind and one without - then combine with sustained.

2) Once you have determined that the load cases are ok, for the case that isn't converging, use the [F2] key to see exactly which restraints aren't converging and why. Wait until the solution starts cycling before you do this. The report will show which restraints aren't converged, and the difference between the last two iterations.

3) While you can use the [F3] key to adjust the friction settings during the solution, I don't like doing this because it makes the solution hard to repeat, especially after 6 months or a year. I prefer to change the friction parameters in the configuration file, and then archive this file with the job.

4) Here are my recommendations, based on the [F2] report:

• If the report shows "sliding - not sliding", the best thing to do is reduce the "friction stiffness". The lower you make this value, the easier it is for the job to converge, but you are making the system very loose, in other words introducing some error into the solution.
• If the report shows "err=xxx", the value "xxx" is the percent change in the normal force on the friction restraint. To help convergence, you need to increase this value. Obviously you don't want to go too high here, what ever you're comfortable with.
• If the report shows "angle=xxx", the value "xxx" is the angle variation between iterations. Again you would increase this value to help convergence. Again you don't want to go too high here.
• If the report shows "open - closed", this is not a friction issue. This is something like the pipe bouncing on a +Y or against a gap. Try changing the gap size or moving the restraint.

You will find other discussions on the topic of friction in this forum. Use the link at the upper right to search this forum.

Some times we use to put Y restrain instead of +Y restrain where we find convergence problem.
Is there any technical difficulty of doing so ?
If load summery show only -ve value we mark only resting support, not hold-down for support detailing.

Regards

Habib