Drawbar tension has been argued ad-nauseum, so instead, how about some physics and 'why everyone can be correct even with vastly different numbers'. First; there is no linear relationship between drawbar tention and the friction generated. The way tapers work is a very large topic, but in a nutshell, the force on the taper (pull on the drawbar in this case) is generating a press fit. If you want to calculate the torque transfer, you would use similar formulae. The inner part of the interface (the adapter) will be considered incompressible and thrown out for the purpose of this discussion. So the amount of interference fit depends primarily on the material of the outer part of the interface (in this case, the spindle itself) and the amount of drawbar force. That may seem contradictory, but the same drawbar force will create different interference with different materials because the material stretch is vastly differnet from one material to the next.
This is where the 'trial and error' would come into play as was mentioned in the last couple of posts. By way of example; my current mill has an X3 spindle which is just soft mild steel. May have been Toyota frames in a past life, who knows. As a guess, I would (generously) give it 50k - 70k yield. My new Mega-Duty 4th axis spindle and the BT30 spindle are 110k yield. A deeply hardened 4000 series steel spindle could easily hit 190k. It is going to take a lot more tension to spread the taper on that last spindle than on my crappy X3 spindle. It would not surprise me if the drawbar tension for a VMC spindle would literally exceed the yield strength of an X3 spindle and permanently deform the taper. A big hole in the nose and the adapter would be up inside the spindle somewhere . . . .LOL!
To address another question: Surface area of the taper is a consideration, of course, but the gorilla is the taper angle.
So before anyone argues that the drawbar alone is retaining the adapter, consider that if the taper angle is very low (i.e. MT, JT, etc) the leverage is great (lots of stretch for not so much force) and the grip is tremendous for that reason. Note that these tapers do NOT need continuous drawbar tension once the initial force has caused the interference fit. It would be a decidedly bad idea to remove the drawbar from R8 for example after the adapter/collet was seated, but the effect of the pseudo press fit is evident when you have to hammer on the holder (thru the drawbar) to free it from the spindle.
In a later post, I will comment on some other topics and show the completed iteration #2 ATC claw . . working perfectly and will be the final design.