One thing to keep in mind is that the rpm must be high enough to allow centrifugal force to keep the starter bendix engaged. If your circuit allows you to vary rpm you could conduct some experiments to see what the minimum engagement rpm is. Then you could generate some EGT vs. time cool down curves at various rpms to determine the optimum rpm for your engine. My guess is that you will have to give it a short burst of higher rpm to engage the bendix, then slow it down somewhat to the cooling rpm.

That would be an interesting experiment. Please post your results if you decide to do that.

Joe