Could someone explain what the use of the artillery tokens in combat is intended to model? I ask because I am trying to figure out why some (attackers) are discarded, while the defenders are returned? Thanks.
The design diary entry referenced and the design notes in the back of the rule explain much of the function of the artillery tokens.
With regard to the defensive vs. offensive use of artillery in particular, when ACW artillery was used offensively, it was by an extended period of firing round after round of shell from a distance at the enemy forces, followed by an infantry attack. This process tended to burn through the firing units' supply of artillery shells, necessitating a re-supply from the supply train: a process that was time consuming and imperfectly executed at best.
When artillery was used defensively, it could be used in two ways: the first was counter-battery against offensive artillery, a tactic that was often thought wasteful but sometimes necessary. (Losses to defensive artillery by offensive bombardment is in part a matter of forcing defensive counter-battery as well as actual damage to the defensive forces.) The most important use of defensive artillery was during the enemy infantry attack, which was for a much shorter time than the offensive bombardment, starting with shell and switching to cannister when the enemy closed. Defensive artillery thus tended to use much less ammunition than offensive use.
This difference in how artillery was used in offensive vs. defensive roles is why the ammunition expenditure rules are different.