If I have eliminated friction at just a few supports in order to converge, and if I think this friction is significant, I would reanalyze specific load cases where I model the friction force by hand or lock the pipe from sliding by my own (entered) perpendicular restraints.
Friction stiffness is the stiffness applied to those (artificial) restraints that the program uses to model the "stickiness" of a support.
If the internal system load perpendicular to the restraint does not exceed the friction load (mu*N), the pipe should stick rather than slip. The program adds two restraints mutually perpendicular to that normal load to prevent this sliding and these restraints have a stiffness as set in the configuration file - this friction stiffness.
Why would this friction stiffness be different than the default stiffness assigned to rigid restraints?
Let's assume you have a series of Y supports with friction along a straight run and this run has some load applied at one end. Assume, too, that the pipe does not slip so friction stiffness is applied in the X & Z directions at all friction supports. In the analysis, that large load will be carried by the closest friction restraint. If mu*N is less than the applied load, the pipe should slip and CAESAR II will replace those friction restraints at that node with mu*N against the slide. The next iteration will do the same at the next friction point down the line. High friction stiffness will isolate the changes to only the next friction support in the line. A lower friction stiffness will allow some of this applied load to "leak" down the line an reduce the number of iterations to a consistent (final) solution. A lower friction stiffness reduces iteration.
However...
Several years ago, when PCs were much slower and iteration became tedious, we shipped a version of CAESAR II with a lower value for this stiffness - 50,000 lbf/in, I believe. Iterations were reduced but we also saw other response that wasn't reasonable. We traced this unusual response to the lower friction stiffness. We returned to the higher default value for friction stiffness in the following release.
I wouldn't change friction stiffness. But that's my choice.
_________________________
Dave Diehl