This MTBF stuff is all very interesting, but if we are talking about theoretical aircraft all powered by engine x (with a certain MTBF), then that MTBF is a constant, and therfore not really relevant.
My maths is a bit dodgy these days, but being a bit of a gambling man, I found this formula:
log(1 - DC)
N = ----------------
log(1 - p)
N=number of trials (in this case number of engines)
DC= degree of certainty
p= probability (in this case MTBF)
So swh, plug in whatever MTBF and number of engines you want into your scientific calculator and let me know what you come up with.