Extracting maximum power from a BLDC

In theory, and somewhat in practice, a motor's power limit is determined only by its ability to (1) keep cool and (2) not fly apart under centrifugal loads. I want to know how to apply theory in order to determine on a case-by-case basis how to push a motor past its published specs.

Assuming I have an appropriate gearing system and a constant torque load, then I can place the peak RPM anywhere I'd like. And assuming some combination of increased voltage-- or even rewinding a motor-- I can achieve any RPM. And assuming I can apply the appropriate cooling through enhanced airflow and perhaps even exotic phase change cooling, then I can keep the motor within operating temperatures. So what are the practical limits in overdriving a motor?

For instance, T-motors sells the new V10L motor, which although spinning the test prop it made 50% less power than I need, it has class-leading heat capacity and weight. It's fair to think that, since the system has a demonstrated 11Nm of torque at 6kRPM, one could run it up to 9kRPM, giving a 50% bump in performance.

(High RPMs aren't as crazy as one might think, LMT motors can run at 50kRPM, and the Dyson cordless vacuum motor reportedly runs at 80kRPM.)

I understand that there are no magic bullets and if all we needed to move a container ship was a tiny 80kRPM motor then we'd already be doing that. I'm not looking for a free lunch, and am instead hoping to expand my understanding of motor manufacturing and performance.

Thoughts?

There are losses associated with rpm. Hysteresis losses and eddy current losses in the stator (and also the rotor to a lesser extent) increase proportionally to rpm and rpm^2 respectively. So more rpm means more heat, which means you can't just increase rpm forever because you're limited by "ability to (1) keep cool."

That motor looks like it has a heck of a lot of pole pairs (guessing 24 pole pairs from counting in the picture?). That would mean that 6krpm is 144kerpm = 2400eHz....

This does not really make sense from a PWM perspective. You won't be driving sinusoidally. In BLDC mode, the phase of the PWM with the rotor will start to be very important.

You need at minimum 6 PWM periods per electrical revolution otherwise it probably won't even spin... so your min PWM frequency will be about 18kHz. Below that and it will probably barely be able to produce any power.

Just a thought... Don't know what pwm frequency or controller you're using but I'd be looking at one that pwms at 36khz+

Edited since i made a divide vs multiply error...