Electrical – Ferrite core through PCB, how to close?

B_max: There is no universal definition about the exact point above which an inductor is called saturated. Many modern ferrites saturate around 300 mT. Useful values tend to vary between the points when the inductor has lost 10 % to 33 % of its original inductance. Here's a practical test useful for finding a reasonable \$I_\rm{max}\$ for a given inductor: You apply a square voltage across the inductor and monitor the current while doing so. You will observe a current waveform that starts from zero in a linear way as soon as the voltage is applied. You can calculate the inductance using \$L = U * \frac{dt}{dI}\$. You will observe an increased rate of rise once you get to the point of saturation. \$I_\rm{max}\$ is reached when the rate of rise just starts to get bigger.

Maximum useful frequency and other parameters: Most interesting parameters (including frequency) have to do with core losses. It's quite hard to find out about them without being able to compare an unknown core to a known sample. When I was designing transformers for switching power supplies, we would test different cores and just see how hot they behaved at different currents, with or without being driven into saturation, at different temperatures and at different frequencies. It was a lot of experimentation and trial and error, and we made samples with different numbers of windings, different air gaps and various ways of layer construction.

If the transformer already has an air gap of ~1.5 mm, an additional glue gap of 0.05 mm will likely make no difference.

If the transformer was glued before you disassemble it, it should be broken right where it was glued, cleaned to bare ferrite and re-glued in the same way. It's likely to keep its performance close to what it was.

If you break the core in some random place and re-glue it, you introduce an additional gap. It depends on the transformer's and the SMPS's design how much it will affect performance.

I have unglued a few transformers putting a knife at some of the glue joints and hitting it with something about 100 g heavy. It succeeded in about half of the cases.