I've taken to setting the friction multiplier to 0 so that I can get a Q&D™ answer for which supports are lifting off, and start taking friction out for these specific locations until I get convergence, adding friction back in, one at a time, until I get as close to a final answer.
However, generally these are at risers, and because of lift-off, they inevitably have a higher vertical load than the others.
But with no friction, this results in no lateral loads, even though this is fictitious, I've had structural types take the no lateral loads case and run with it. I've also had other structural types average out the lateral loads and run with that. Both are inaccurate, so long as the vertical load at those particular supports are much higher than other "average" supports.