Its tricky as you break the loop with your XBP2 setup, so the loading of the input and
output of the freq determining network is not accurate. And you have not shown the
parasitics for the bipolar, like miller C and varies junction C.
That being said you can do an approximation with some assumptions :
1) Look at just the network consisting of the xtal and its 12 pF loading caps. In
simulator drive input to it with a current source, and using hybrid pi model
approximation calculate Zin of the transistor and use that for loading across the
output 12 pF. Then look at sim plots for when the phase shift becomes 180 degrees.
OR
2) Use signal flow graph (SFG) approach to evaluate the entire circuit input to xtal to
output at collector of transistor, and that will get you in ballpark. Here you
would look for 360 degrees of phase shift. Tedious but SFG makes it a lot
easier for high order networks. Search web and you can find GUI utilities
to construct the graph and its tool will generate the T(s) for you.
Regards, Dana.