>Efficiency is usually somewhat higher at high voltages, because the ohmic losses (for given wiring and inverter) get lower for a given power output.
If motor is wound for that voltage. But I think it is not the case here.
There are two things in motor that are important. Torque and speed.
Torque is a function of current, which is limited by controller. If I keep 15A current limit, it would accelerate with exactly the same rate, until difference between BEMF and battery voltage drops to the point it can't push said current.
Speed is just a function of voltage.
When it comes to the current, losses in the motor are exactly the same (since both copper and iron losses are function of current iirc), but losses in the battery would be lower.
When it comes to the losses at 25 km/h... I dunno. I think that controller that works as a commutator (i.e. duty cycle is 100%) losses are smaller than when it works as synchronous buck converter (i.e. duty cycle <100%). And I think that buck converters (e.g. pic related) are more efficient when Vin is close to Vout. Which again, is better, since BEMF at 25 km/h is probably very close to supply voltage.
>Ultimately these are just current spikes in the FETs, so you can probably calculate the losses at certain voltages if you have the Kv rating for the motor.
KV of the motor is approx 1.0602 km/h per volt. A bit higher probably, because I've adjusted speedometer so it is a bit more pessimistic than reality.
But then idk the inductance of motor or frequency of the controller. I don't even know the resistance of the motor.
So its kinda hard to calculate absolute numbers, and relative numbers - idk either