I'm reading the procurement documents for the R211 railcar contract, which is for electric multiple unit subway cars designed for rapid transit applications in New York City. The railcar's propulsion is specified to use AC induction motors powered from PWM inverters with IGBTs.
The specification related to propulsion motor overcurrent is given as follows:
10.3.20.2 Over-temperature protection shall be provided as follows:
a) If the winding temperature rises 27°F above (15°C above) the operating class temperature limit, the control unit shall decrease the motor duty cycle by removing the dynamic braking and converting to frictional braking, until the temperature falls below the operating class temperature.
b) If the winding temperature rises above its design class operating temperature limit, the control unit shall remove power to the affected truck, until the temperature falls below the operating class temperature. The condition shall be reported and recorded by the MDS.
c) Motor temperature shall be calculated using a thermal model of the motor and the power that the motor is handling.
d) The measurement of winding temperatures using embedded sensors is not permitted.
Points c) and d) stand out to me as a little bit strange, since the motor is expected to be modeled with its loads, instead of directly measuring the temperature. The only reasoning I can determine for this is that it protects against the failure of the thermal sensor.
However, it's not clear to me why this failure more is harder to guard against, than (for example) the failure of current sensors needed to track the motor's power, or an obstruction of the normal cooling process by dust or debris. Is there a common technical reason why motor thermal protection is specified to use only modeled heating rather than measured heating or a combination of safeguards?