In order to determine the reason for the discrepancy in the magnitude of the shaft torque
negative peak, all of the data must be examined.
The most accurate measurement is the speed measurement. Since a 1024 ppr encoder is used
for the speed acquisition, the speed measurement has very high noise immunity and is
therefore the most reliable signal. The oscillation seen in the speed waveform is not noise, it
is in fact a real oscillation due to the pressure oscillations generated in the water brake load,
as previously mentioned.
The current measurement uses a hall effect current transducer rated for 2000A peak and was
recently calibrated to NIST standards. Since the magnitude of the peak current is
significantly less than 2000A, the current measurement is also assumed reliable.
The torque transducer measurement is probably the most questionable measurement,
however, these tests use a state of the art torque transducer that guarantees transient
performance within a bandwidth of 1.1kHz and high noise immunity . The rating of the
torque transducer is 100,000 in-lb, with a x2 overload rating. The magnitude of the peak
torque measured for the 50 hp tests is less than 50,000 in-lb and the highest frequency
component measured during the negative torque transient is less than 800 Hz. The shaft
torque measurement is therefore assumed to be reliable. A closer look at the speed data
reinforces this assumption.
The simulated speed waveform agrees well with the measured speed waveform, until the
stator voltage is re-applied. At this point, the simulated speed waveforms predicts the shaft
speed falling to about 1420 rpm, whereas the measured data measures a shaft speed of about
1350 rpm at the same instant in time. This is a difference of 70 rpm. In order for the
simulated speed to fall an additional 70 rpm, a much larger electromagnetic torque must be
The magnitude of the peak current measured in Phase C during the experimental testing is
775 A. The magnitude of the peak simulated current in Phase C is only 575 A. This is a
difference of 200 A. If the simulated current peak increased to the value of the measured