Determining peak of normal distributed data

Can you try abstracting your question a bit as well as putting it in reprex?

At least for me it is not clear what exactly you want to do and especially since there are a lot of moving parts and your set-up is very specific.