[RPG] How do spell effects stack when determining the DC to disbelieve an illusion

Encouraging inventiveness is easy. You've already done it, in fact: You simply have to reward it by allowing it to (occasionally) succeed.

Avoiding breaking the game is also easy: You simply disallow anything that lets players to bypass the obstacles presented by the game with less effort than it would take to tackle those obstacles in a more traditional way.

Obviously, these two goals are hard to reconcile, but I can suggest a few things that might help.

First, let your players have the benefit of their invention, but remember to account for the costs: In your example, your alchemist has developed a devastating weapon, but one which eats through very expensive ammunition at a disturbingly rapid rate. Once the disparity between party member wealth levels grows a bit, he might start to worry. Also, given that it is a devastating weapon, even if it is expensive, ask yourself why other people don't use it. I imagine it would take just one orc with the Improved Sunder feat to make your player realise that there are pitfalls involved in waving acid vials over his own head.

Second, remember to include at least a few situations in which the invention does not work, to keep your players on their toes (and encourage future inventions): There's no end of creatures immune to acid and fire, so let them crop up occasionally - and your alchemist would be in strife if away from civilisation for an extended period.

Finally, don't be afraid to condtradict previous rulings if you can come up with a good reason for it. Yes, you should avoid it most of the time, but inventions are a special case, since by definition they're trying something new and unknown: Invention is inextricably paired with discovery. Be tactful, and start by saying "There's something I didn't think of that's kind of relevant to that idea you came up with," but be firm, and try and provide a reasonable in-game justification for it. "Your acid net? The ropes that make it up only have about two hit points, so it should be disintegrating every time you use it. Also, how have you been protecting yourself from the splash?"

To the best of my knowledge, the issue is never directly addressed. The rules, on a quick scan, appear the same as in 3.5, where this issue has been debated some as well.

The long and short of the argument is this: the statement in shadow evocation et al. is that someone who knows that it is fake does not need to save. Nothing says he does not save or may not save, just that he doesn’t have to. This is taken to mean that this is optional, and effectively someone in this position has the option of automatically succeeding on his save the same way you typically have the option of failing any save. You may, according to this logic, choose not to automatically succeed, and then, since you are now attempting a saving throw, choose to automatically fail.

Strict-RAW, this seems most accurate, though it definitely takes a few steps to get there and it’s clearly not written out explicitly. Still, the language, whether it was intended to be or not, is precise: it waives a requirement to save, it does not add a requirement to not-save.

Whether or not you should rule this way in your game is more dubious. Shadow evocation et al. are rather useful, particularly for this feature. In 3.5, greater shadow evocation was typically used to cover the loss of contingency due to the banning of Evocation as a specialist wizard. In Pathfinder, this is less of an issue (since banning is no longer so absolute anyway). Most of the time, shadow evocation et al. are most useful when the drawbacks of using them (the Will save, the quasi-reality) don’t actually affect the functioning of the spell, which is precisely in this case: buffing. Ultimately, it becomes yet another powerful and flexible tool in the wizard’s toolbox, and he’s already got a ton of those. Shadow evocation et al. are’t the most powerful of them, but maybe it’s worthwhile to you to start paring down options where you can.