This is a "feature" of Logic.

DG

pardon me if I am stating anything that is already obvious.

when using multi-timbral instruments in Logic:

MIDI CC automation on a region isn't quite the same as automation on the track on which the information is recorded - as you can move automation from the region to the track and vice-versa. But its not a complete transform either way. You can lose automation.

So performing a volume ride on the midi track (or instrument channel of a multi-timbral) affects all those midi tracks for that region.

My solution was to only perform MIDI CC automation on the region and not even touch the volume of the track (hope thats clear, I am not using maybe the right terminology)

some CC's that might be useful

1 - mod

2 - breath controller (which I mapped to velocity crossfader in VSL)

5 - portamento time (esp. useful in EW PLAY for legato / portamento control)

7 - Volume

11 - Expression

And use the Hyperdraw functions in logic as well as the CC automation view in the piano roll. between key switching and mod wheel control, I get access to all the different samples in each instrument channel. 

From what I have read, most people in these forums seem to advise setting overall channel instrument volume using CC7, and then putting in specific cresc / decresc with CC11 (or in my case, I am using CC2 to activate the velocity crossfader with VSL as it sounds good to me)

in addition, with multi-timbral instruments, you can use it as stereo or multi-output, and have aux channels activated for each output or pairs of outputs that the instrument has. 

i.e. - create new track, choose EXS24, choose multioutput. 

In the mixer, look at the channel you just created, at the bottom near the channel name should be a small + sign.

click that and for each click, a new aux channel will be created for each output of the instrument. 

Once you have a bunch of aux channels, then each will handle the output of corresponding midi channels for the same instrument. And each aux can have pre and post fader effects on it, and volume automation drawn on the aux as well. Each aux will have individual control. 

Using VSL and VE PRO, I have found this way of working to make sense to me. Using  EW PLAY QLSO, it's a little less useful so I don't bother with the multi aux creation, but still make the instrument multi-timbral and record region automation for all those CC's

Hope this rambling story helps you.