The tracking mode is there so that when you transfer control from one PID loop to another, or when you transfer from manual control to PID control, there is no big transient, i.e., you achieve bumpless control transfer. This example explains bumpless control transfer from manual operation to PID control.
As I mentioned, another common scenario is transfer from one PID loop to another. For example, say a sensor in your car engine fails. In that case the controller switches from normal mode to a back-up mode. Assume that these two control modes are implemented by two different PID controllers (in normal mode you are a bit faster, and in backup mode you are slower and more conservative). When this transition changes you want the fuel flow that is commanded to the engine not to jump suddenly when the failure happens, but to change gradually. This can be accomplished using the tracking mode.
Hope this helps
