Similar to you, but slightly more compact.
More or less in your notation:
WWWWWW ~ ~ ~ WWWWW.
W (for >..
W X... ~ ~ ~ ....W.
W the >..
W X... ~ ~ ~ ....W.
W max >..
W X... ~ ~ ~ ....W.
W length >..
W X... ~ ~ ~ ....W.
W needed) >..
WWWWWW ~ ~ ~ WWWWW.
^ ^
dowm to bolt stores-' |
|
main corridor---'
The space behind the target is not necessary for bolts fired at the target itself (all destroyed, anyway) but bad misses that stray across shooting lanes might have hit the wall behind, and this way it lets them land in the lower area as per less/more inaccurate misses. (Not entirely sure if "just bad enough to hit adjacent target" misses destroys the bolt. Probably. It should be testable, even without gluing one's eyes to the screen, by having an inactive range in the midst of the set and seeing if destroyed bolt icons eventually appear on the ground in front of it, fired from adjacent (or further!) lanes.)
Whether I have a lower-level bolt stockpile defined all the way from the access stairs to the archery target end, or not, differs. If I don't put one all the way from end to end, I tend to put a smaller one at the stair end, giving haulage dwarfs the task of getting loose fallen bolts to the end more quickly accessed by the archers needing more ammo. Alternatively, putting it at (or fully extending to) the target-end might count them as being stockpiled already, if I was bothered about hauling times or distractions from some other moving job.
I don't actually see why I couldn't make the target walkway into a line of stairs-down floors (counterparted with stairs-up on Z-1, of course), or keep some of the ramps created on Z-1 by the initial "channel away" part, rather than remove them all to allow stockpile existence on those tiles) to allow quicker travel to all places the missed bolts land, but I never have. Probably more an aesthetic choice. As is the fact that I smooth all walkways (even under the spot the target will be built) and walls. At least on the archery level, if not also on the drop-level.
Thus my design is comparably intensive in labour (shared 'channel' walls save miner-time, but the engravers spend more time in there and the Architect might not be called in until later than he really could be), depending on how you weight the various efforts and if you use micromanagement burrows or exploit the intrinsic job-prioritisations to get a better critical-path workflow in your creation.