If you're specifically looking at what required frequencies of folding you need for a profitable bluff, then you would look at what frequency your opponents/population are folding for that node in your database. Additionally, it might be helpful to look at solved sims for what hands are bluffing/calling and the frequencies/hands for each.

I think you're going about this a bit wrong. You are deciding your betsizing based on the bluffs. This is the opposite of what you actually should do.

Betsize is driven by our value hands and only our value hands.
Think about it. Your opponent calling the proper amount of times makes it so that your bluffs are 0ev. If you size higher with bluffs and they fold more, still 0EV. If you size down and they fold less often, still 0EV.
So the only hands that profit from different betsizes are our value hands.

I think the reason you’re seeing sizing that’s different from what you’d expect is because you’re approaching sizing from the wrong perspective. You’re trying to stay balanced. This means your sizing as a bluff needs to follow a logic that works for your value hands, too. If you have AKs and the flop is K72r, you will check or bet small to get value from worse, because there is very little your opponents can continue with. For the same reason, you can bet smaller when bluffing on these boards. You don’t need to bet a large amount to have your opponent fold on dry, static flops. However on wet flops, you need to use a bigger sizing to deny equity, and it’s the same line you’d take if you have a value hand on a board like JT9 with a suit.

This also makes bluffing more profitable and more likely to get through on dry boards vs wet ones.