While modeling a 2 pole motor for 250Hz, we found out that at frequencies above 80Hz there is little to no difference in efficiency between a magnetizable or non-magnetizable shaft. We found this odd, because the field lines go through the shaft in either configuration. So we put a motor with and without a magnetizable shaft on the test bench and tested it between 0 and 135Hz. The results were compliant with the model, but not able to explain the phenomenon.
There is a relationship between applied voltage, magnetic flux, and frequency to the motor. It can be expressed as:
(applied volt) = (magnetic flux) * (frequency)
With the modeling you're doing for elevated frequency designs, I can't see you also increasing the applied line voltage by 5x (for nominal 50 Hz design) or 4x (for nominal 60 Hz design) to correspond with your 250 Hz operating point. Therefore, the amount of magnetic flux generated is going to go WAAYYY down as the frequency runs up into the 100+ Hz range.
If the overall flux is going down, it won't necessarily have to traverse the shaft material, since the lines of flux can reasonably pass through the lamination material (possibly resulting in similar flux densities for that short section of the circuit as when operating at more "normal" frequency ranges and passing flux through the shaft).
We have recently started to look at lower frequency with the need to replace ancient 25 Hz motors. Testing has revealed that we can use a standard 2 hp 2 pole motor to replace a 0.5 hp 25 Hz motor without excessive heating. The actual results were better than computer model and it looks like we have similar efficiency to the old motors without any modifications.