I’ve taken the basic premise of what @Microfone9986 has outlined here…
…and programmed a basic flight plan directly into the FMC of the 747 while sitting on the runway. I’m not bothering with ATC, because I want to see how the FMS behaves once I program an approach shortly after takeoff.
The flight plan departs KSFO, follows various waypoints in the SF Bay Area and returns to KSFO in a loop just as he has done with KJFK.
What is interesting is (without all the pressure of ATC during a “normal” airliner IFR flight as setup in the World Map and directed via ATC) being able to actually see how the FMS misbehaves after the approach has been assigned. Since the waypoints are all so close together (rather than spread out by massive distances in a “normal” flight plan), you can easily see what the aircraft is up to when it does the strange turn away from the assigned Transition waypoint.
In my case, I can see that even though the top waypoint on the LEGS page is the Transition, the FMS has, instead, directed the aircraft to the final waypoint I’d programmed into the FMC, as seen on the final page of RTE.
I want to experiment a bit more and report if this is the consistent behavior across the board, or just in this particular case.
Anyhow, it’s giving me some insight into why we see such erratic/unexpected/unintended behavior with the autopilot once we assign an approach.