The intent is to supply a greater stiffness where there is greater friction. This appraoch does that.

This works quite well for calculating natural frequencies. I usually use 1000 as my stiffness factor for friction but find that a value of 200 works about the same.

If you are looking for that friction to carry some dynamic load (as opposed to simply calculating natural frequencies) I guess you may want to have specific control at each support but I don't think you really know what's going to be happening there anyway.

If you want to specify unique stiffnesses at each sliding support, why not model this friction restraint as individual snubbers?