Proper Managed VDI Pool VM Configuration for UPS Shutdown?

I am preparing to set up our Server 2012 Datacenter machine to shut down cleanly using a UPS after power-failure.  Shutting down is easy, but my goal is to make the shutdown cleaner.

When a Managed VDI Pool (running Windows 7 Enterprise) is created, the default Automatic Stop Action for the VMs in the pool is Save.  This would make sense if everything would work right upon boot, but that doesn't seem to happen.

I haven't tested this extensively, but in the experience I have had, if I shut down the server with users connected to VDI, their sessions are saved, but the VMs are no  longer reserverd for their use on boot.  To be more specific, here is a an example of a chain of events I actually experienced:

1. User 1 logs into VDI Pool and is assigned VM A1
2. I shut down the server and VM A1 is moved to a saved state
3. The server is booted back up and VM A1 stays in a saved state
4. User 2 logs into VDI Pool and is assigned VM A1
5. User 1 logs into VDI Pool and is reconnected to their session, disconnecting user 2
6. User 2 reconnects and is reconnected to their session, disconnecting user 1
7. Repeat steps 5-6 until someone contacts me and I have one of the users log off

This obviously isn't ideal, and it doesn't seem like the appropriate behavior (saving the user's session over a power failure would be great, but having users fighting over access to the same VM because something in RDS doesn't know the previous session still exists is definitely problematic).  That having been said, I would like to know what the best practice is for shutting down a RDVH server with a Managed VDI Pool when users are connected.  Obviously it would be ideal to have them log off, but in a power failure situation, they are no longer connected in order to log off, and there isn't likely to be someone manning (or available to man) the server in order to force logoffs before the server is shut down.

To be clear, I am working with a production system, so I haven't made any changes in an attempt to resolve this because I don't want to break anything.  That having been said, if this were a test system, my thinking is that I could set the VMs to shutdown on power outage and that would force the users to log out, which might cause the VMs to be restored to their managed state, but with the server shutting down, I'm not sure that would happen, and the restore process could be interrupted if it did.

That having been said, can anyone tell me how they've dealt with power outages while using VDI Managed Pools, point me toward documentation that covers the same scenario, or even point me toward documentation that indicates what changes should and should not be made to VMs being managed by RDS?