(bio)equivalence tests: acceptable difference

As you may know, (bio)equivalence tests are tests that are able to show equivalence between two compared groups, other than the more common tests that show a difference between them. The most basic form of equivalence testing is the two one-sided test (TOST) procedure.
Typically in these kinds of tests, one needs to predefine an acceptable difference (delta) to be able to make decisions. From the confidence interval (usually 90%) around this value, one can see if two compared methods can be considered (bio)equivalent or not.
The choice of this value depends on many factors and is therefor often a difficult one, certainly when no historical data is available.

I'm looking for some general guidelines and recommendations in the determination of this acceptable difference, because I can't find much literature about this. Is there a way to calculate this value, and how can various factors influence these calculations?

Thanks in advance,