1. Even though there is no lift off, there is load redistribution. Think of it like jacking up your car, the four tires remain in contact with the ground for the first few cranks but you know their loads are changing. Friction is discussed below.
2. If you would analyze sustained loads in a model where all +Y's that lift off (in operation) are removed from the analysis, these points will sag. That negative displacement should match the your L5 deflections.

Now friction...
In my opinion, this sustained equals operating minus expansion approach works quite well for lift off. If you gain confidence in this approach and you wish to evaluate sustained stress for all support configurations, it will speed your analysis as you will not have to separately re-analyze sustained stresses for the various operating cases where supports lift off. But friction complicates things. The expansion that you are subtracting was created without deadweight and without deadweight, the friction components in the expansion set are probably underdeveloped. So those (larger) friction effects found in the operating state are not properly negated by subtracting the punier effects of the expansion set.
Again, my opinion.