Model verification with prototype motor temperature measurements

Simulation model was verified in the laboratory with the prototype motor connected to an ABB ACS600 frequency converter. The machine was loaded with a Vibro-Meter torque transducer, which is an eddy-current brake connected to the control and measurement equipment. The motor temperatures were measured by using Pt100 temperature sensors, which measure the variation of the resistance due to temperature. Altogether 14 sensors were assembled; one for each phase at both end windings, three for the frame (one in the centre and two at machine ends), and similarly three for the stator yoke and two at the end caps air. Sensors were attached by using heat conducting Araldite glue. Temperature signals were received from the sensors by using Fluke Hydra and a laptop PC equipped with Labview software. The same software was used in obtaining motor currents and voltages from the power analyzer. A minor problem in using Fluke Hydra was its poor suitability for fast temperature transients. When temperatures in five channels were read (e.g. three phases, the frame, and the yoke), the minimum sampling period was approx. 15 s.

Two transient and two static measurements were carried out to verify the validity of the model.

Both transient measurements were carried out in overloading conditions, as the overloading of the motor is very common in servo drives. Also a slight adjustment of thermal network parameters

was carried out according to the measurements, although the parameters obtained from pre-calculations and low-frequency tests were surprisingly close to the final values. The biggest source of uncertainty was the radiation resistance from the frame to ambient. Although the emissivity of the matt black frame can be roughly estimated to be approximately 0.8−0.9, it is impossible to estimate the ambient absorptivity even roughly. Therefore, the final value for the radiation thermal resistance was obtained by comparing the simulation data to the measurement data. This value gives a 20 % share of the heat dissipated through radiation, when the motor operates near the maximum winding temperature (where Tframe ≈ 100°C).

In first transient measurement, the machine was loaded with the 2.3-fold overload for 15 min, after which the motor was disconnected from both the load and the inverter (zero losses), and the cooling period of 15 min followed. Temperatures were measured within 15 s intervals from the frame, the yoke, and three phases at end windings. The measurement and simulation data is shown in Fig. 4.10.

20.0 40.0 60.0 80.0 100.0 120.0 140.0

0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 1800.0

Time [s]

Temperature [°C]

Coil_sim Frame_sim yoke_sim Coil_meas Frame_meas yoke_meas

Figure 4.10. Comparison of simulated and measured data, where the cycle consists of 15 min with 2.3-fold overload and 15 min cooling period (no losses). The agreement is very good, although there is a slight error at winding temperatures (4 °C at maximum). The measured temperature of the stator winding is the average value of the three phases.

The second transient measurement was the repeating cycle, where the machine was loaded with the 2.3-fold overload torque for 2 min, which was followed by the 1 min cooling period, and the cycle was repeated again. This measurement is shown in Fig. 4.11.

20 40 60 80 100 120 140

0 200 400 600 800 1000 1200 1400 1600

Time [s]

Temperature [°C]

End winding (Measured) Frame (Measured) Yoke (Measured) End winding (Simulated) Yoke (Simulated) Frame (simulated)

Figure 4.11. Measured and simulated temperatures, when the three minute repeating cycle contains 2 min loading with the 2.3-fold overload and 1 min off-period (zero losses).

As can be seen in Figs. 4.10 and 4.11, the accuracy in the end winding temperature calculations during over loadings is relatively good. At the rated frequency, the frequency converter did not cause any errors to the measured temperatures, as there clearly is no change in temperature after the frequency converter switch-off in Figs. 4.10 and 4.11. However, there seemed to be certain frequencies below the rated one at which the electromagnetic interference of the frequency converter caused significant fluctuations to the temperatures. Temperature sensors were attached by using glue, which has some small thermal capacitance that could slightly smooth out the measured temperatures. Temperature transients in Figs. 4.10 and 4.11 are high, and during such a high transient, the temperature rise rate is to a large degree limited by thermal capacitances that can be calculated accurately, if the complete dimensional and material data of the motor is available. This partially explains the good accuracy in Figs. 4.10 and 4.11. Therefore, also the steady-state accuracy of the model was verified by running the motor with a constant load in two operating points, until the thermal equilibrium was reached. Figures 4.12 and 4.13 show the temperature rise of the motor when loaded with a constant load until the thermal equilibrium is reached (thermal equilibrium was assumed when the temperature rise was below 0.1 °C in 5 min).

In Fig. 4.12 the motor is operated at the rated torque and speed, and in Fig. 4.13 both the speed and the torque are 50 % of the rated value. The measurement in Fig. 4.13 was carried out on a hot day in July, when the laboratory temperature was almost 30 °C, which explains the difference between the initial temperatures.

20 40 60 80 100 120 140

0 50 100 150 200 250

Time [min]

Temperature [°C]

End winding Stator yoke

Frame

Dashed line: Measured temperature Solid line: Simulated temperature

Figure 4.12. Measured (dotted lines) and simulated (solid lines) temperatures of the prototype motor, when it was operating at the rated point.

20 30 40 50 60 70 80 90 100 110

0 50 100 150 200 250

Time [min]

Temperature [°C]

End winding Stator yoke

Frame

Solid lines: Simulated temperatures Dashed lines: Measured temperatures

Figure 4.13. Measured (dotted lines) and simulated (solid lines) temperatures of the prototype motor, when it was operating at 50 % of the rated torque and speed.