I agree with Dan's statement that "you can't take credit for friction". We've always advocated to err on the side of caution, especially with friction. Most people can't agree on a friction coefficient, let alone sliding versus static versus dynamic effects.

This is why we added the "friction multiplier" in the static load case setup. You can run:
a) ope w/ expected friction
b) ope w/no friction
c) ope w/ +25% friction
d) ope w/ -25% friction

All without changing the model input. I'd use the restraint summary and take the worst case - individually for each restraint.