INSTALLATIONS FOR STUDYING HEAT-TRANSFER PARAMETERS IN LOW-POWER ELECTRICAL MACHINES

D. I. Sannikov, R. Ya. Klyayn, M. A. Sannikova

(Presented by the scientific seminar of the Departments of Electrical Machines and General Electrical Engineering)

In the experimental study of the heating and cooling of electrical machines, various models, mock-ups, and other special installations for studying individual elements of the thermal system are successfully used. These make it possible to substantially simplify the experiment, limit the number of influencing factors, and take their influence into account more accurately than in experiments on actual machines.

At the Department of Electrical Machines of TPI, the following have been created:

1) a mock-up of a low-power DC machine for measuring the heat-transfer coefficients of the armature surface,

2) an installation for measuring the thermal conductivity of winding insulation.

The study of heat transfer and thermal conductivity on models pursues a twofold aim: first, to refine methods of thermal calculation; second, to develop the designs of newly designed machines. Existing methods for the thermal calculation of low-power machines give a considerable error, primarily because they are based on insufficiently complete experimental data. For this reason, for example, the armature heat-transfer coefficient is determined, as a rule, only as a function of the peripheral speed, while the remaining factors—such as the dimensions of the armature, the structure of its surface, the configuration of the interpole channels, etc., which undoubtedly affect heat transfer—are not taken into account. This makes it difficult to choose designs and basic dimensions when designing highly utilized machines. Reference data on the thermal conductivity of insulation are also quite insufficient.

Measuring the armature heat-transfer coefficient on a mock-up offers the following advantages in comparison with tests carried out on actual machines:

1. Lower labor input in manufacturing the mock-up in comparison with the machine; relative simplicity of changing dimensions during experiments; the possibility of reducing the heat capacity in order to shorten the heating time during the experiment.

2. Simplicity of measurements, which consist in determining the power of the heaters and the surface temperature; no need for $\alpha$-sensors with complex measuring circuits.

3. Higher accuracy in measuring the heat flux in comparison with an actual machine, since steel losses and additional losses are practically absent; ease of regulating the magnitude of the heat flux by means of heaters.

4. Greater accuracy in measuring the average surface temperature by means of thermocouples, owing to the greater uniformity of heating.

1. The possibility of separately investigating the active part, the end parts, and the commutator, provided that they are thermally insulated from one another.

The design of the mock-up is shown in Fig. 1. The mock-up is made on the basis of a series P motor. The active part of the armature is represented by a hollow steel cylinder 1. Grooves are milled on its surface, imitating the slot teeth of a real armature. A heater 2 made of nichrome wire, wound toroidally on an annular frame, is pressed tightly against the inner surface.

Fig. 1

The end parts are represented in the form of two heaters 3, fastened on steel winding holders by means of bands. The commutator mock-up is made in the form of a copper cylinder 4 with a smooth surface, inside which a heater 5 is built in.

The heaters are supplied with direct current through contact rings 6; the commutator mock-up is also used as one of the rings. Thermal-insulating gaskets 7 are placed between the individual parts of the mock-up; with the slight temperature difference of these parts maintained during the experiment, they practically eliminate heat exchange between them. The design of the mock-up makes it possible easily to replace parts in order to change dimensions. The pole windings are bifilar, and the inductor is demagnetized. The model is driven in rotation by an auxiliary motor with adjustable speed; provision is made for regulating the flow rate of the cooling air.

Copper-constantan thermocouples 8 are used to measure temperature. Nine thermocouples are built into the active part, three on each generator of the cylinder, spaced 120° apart. In each end part, six thermocouples are installed under the band, on the mean circumference at intervals of 60°. There are three thermocouples in the commutator. The ends of the thermocouples are brought out through a channel in the shaft to an attached current collector. The braking torque is measured by a compensator. The character of the temperature distribution in one of the regimes is presented in Table 1. The average temperature rise of the active part is 50.0°C,

Table 1

Temperature distribution in the armature mock-up of the P32 motor
at 1500 rpm

Ambient temperature: 25.5°C.

of the end-winding parts is 49.2°C, and of the commutator, 49.1°C. Thus, owing to the correct selection of the heater powers, the mock-up is heated uniformly. The root-mean-square scatter of temperatures for the active part is \(\sigma_a = 1.73^\circ\mathrm{C}\), or 3.5%; for the end-winding parts, \(\sigma_l = 3.8^\circ\mathrm{C}\), or 7.8%. The mean absolute value of the scatter is, respectively, \(\eta_a = 1.42^\circ\mathrm{C}\), \(\eta_l = 3.1^\circ\mathrm{C}\). In both cases

\[ \frac{\sigma}{\eta} = 1.22 \simeq \sqrt{\frac{\pi}{2}}, \]

therefore the law of distribution of the scatter may be regarded as normal. The root-mean-square error in determining the mean temperature rise is:

for the active part: 1.2%,

for the end-winding parts: 2.2%.

If one takes into account the error introduced by the current collector, measuring instruments, and thermometers, as well as the scatter in thermocouple characteristics, possible heat leaks in the mock-up, and the inaccuracy in determining the area of the cooling surfaces, then the root-mean-square error in determining the heat-transfer coefficients is estimated at 2.5–3.0%.

The value of the heat-transfer coefficient for each of the armature parts is determined from the formula

\[ \alpha = \frac{P}{S_{\mathrm{cool}} \, \theta'}, \]

where \(P\) is the power of the corresponding heater,

\(S_{\mathrm{cool}}\) is the cooling surface,

\[ \theta' = \theta - \frac{1}{2}\Delta \theta_f, \]

\(\theta\) is the mean surface temperature,

\(\Delta \theta_f = t_{\mathrm{out}} - t_{\mathrm{amb}}\) is the heating of the air in the machine.

Some measurement results are given in Table 2.

Table 2

Determination of the heat-transfer coefficients of the armature of the П32 motor
(armature diameter — 120 mm)

The installation for measuring the thermal conductivity of insulation is shown in Fig. 2; it is intended for the investigation of various sheet insulating materials and combined multilayer insulation of slots, coil holders, coils, etc. In view of the small size of the specimens tested \((90 \times 90\ \text{mm})\), the installation is suitable for studying the insulation structures of low-power machines.

Fig. 2

The sheet of insulation being tested, 1, is placed between two brass disks, 2, each of which has four built-in thermocouples. The upper disk is screwed to cover 3, which has cooling fins; the cover is blown by a fan. Under the lower disk is the main heater, 4, consisting of a nichrome coil in an asbestos-cement housing. There is a shielding ring, 5, provided with a heater and four thermocouples to monitor equality of temperatures with lower disk 2. It is mounted on an elastic gasket, 6, made of silicon-organic rubber and is fastened by screws to plate 7. The latter is separated from base 8 and the fastening studs by textolite bushings and washers, 9, in order to reduce heat leakage. Lower shielding heater 10 is separated from the main heater by an insulating spacer, in which 8 thermocouples are arranged in pairs to monitor equality of temperatures between the heaters. All cavities of the installation are filled with glass wool.

The use of air cooling instead of the water cooling usually employed in such cases considerably simplifies the installation.

The temperature of the tested insulation specimen and the temperature difference across it are within the limits characteristic of the operating conditions of insulation in electrical machines. These quantities can be regulated by changing the power of the main heater and the cooling conditions of the cover.

To improve the thermal contact of the insulation being tested with the disks, their surface is coated with a thin layer of paraffin.

The heating time of the installation from the cold state to the steady-state temperature is about three hours. As an example, Table 3 gives the results of measurements performed on a mica-tape specimen of thickness \(=0.35\) mm.

Table 3

As can be seen from the table, the scatter of the thermocouple readings at each position is small, especially for disks 2, which is most important for measurement accuracy; in this case the thermocouples of each position are most conveniently connected into a thermopile; the powers of the side and lower shielding heaters are adjusted to the minimum difference in the readings of the control thermocouples. The power of the main heater is \(P=11.0\) W. The temperature rises of the lower disk are \(47.4^{\circ}\mathrm{C}\), and of the upper disk \(32.1^{\circ}\mathrm{C}\).

The temperature difference across the tested specimen is \(\Delta\theta = 47.4 - 32.1 = 15.3^{\circ}\mathrm{C}\). The area of the lower disk is \(S=28.3\ \mathrm{cm}\).

Thus, the specific thermal conductivity for this sample is

\[ \lambda = \frac{P \cdot \delta \cdot 10^{-1}}{S \cdot \Delta \theta} = 0.00089 \ \text{W}/\text{cm deg}. \]

The error in determining \(\lambda\) for this apparatus is approximately 5%. The second version of the apparatus has heaters in metal housings, owing to which the nonuniformity of the temperature field between the heaters is substantially reduced, and the accuracy of adjusting the shielding heaters and of measuring \(\lambda\) as a whole is increased.

Thus, the installations described are comparatively simple, convenient to use, and provide sufficient accuracy of thermophysical measurements.

REFERENCES

1. G. M. Kondrat'ev. Thermal Measurements. Mashgiz, 1957.