IMPROVING THE RELIABILITY OF RANDOM-WOUND WINDINGS OF ASYNCHRONOUS ELECTRIC MOTORS

Yu. P. Pokholkov, E. K. Strelbitsky

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

A quantitative assessment of the reliability of random-wound windings of asynchronous electric motors showed that, in developing measures to improve reliability, the main attention should be paid to turn-to-turn insulation [1].

Analysis of the causes of breakdown of turn-to-turn insulation can be carried out with the aid of a mathematical model of the reliability of turn-to-turn insulation [2].

\[ R_{\mathrm{v}}=\left[1-q'(U_{\mathrm{k}})-q''(U_{\mathrm{k}})\right]^{Na}, \tag{1} \]

where \(R_{\mathrm{v}}\) is the reliability of the turn-to-turn insulation;

\(q'(U_{\mathrm{k}})\) is the probability of breakdown of a pair of conductors of length \(\dfrac{l}{a}\) that, before impregnation, had damage of the bared type;

\(q''(U_{\mathrm{k}})\) is the probability of breakdown of a pair of conductors of length \(\dfrac{l}{a}\) that, before impregnation, did not have similar damage;

\(a\) is the number of damaged samples of length \(l\);

\(U_{\mathrm{k}}\) is the voltage falling on an individual coil of the winding, taking into account the occurrence of commutation overvoltages.

The probabilities \(q'(U_{\mathrm{k}})\) and \(q''(U_{\mathrm{k}})\) are related to the winding design, the quality of the insulating materials, and the impregnating compound.

\[ q'(U_{\mathrm{k}})=\frac{a}{n} \left[ F\left(-\frac{\overline{U}}{\sigma}\right) \right]^2 \lambda \int_{0}^{U_{\mathrm{k}}} e^{-\lambda U} \left(1-\frac{U}{U_{\mathrm{k}}}\right)^2\,dU, \tag{2} \]

\[ q''(U_{\mathrm{k}})=\frac{a}{n} \exp\left[\lambda(2\overline{U}+\lambda\sigma^2)\right]\lambda \int_{0}^{U_{\mathrm{k}}} e^{-\lambda U} F\left(\frac{U-2\overline{U}-2\lambda\sigma^2}{\sqrt{2}\sigma}\right) \left(1-\frac{U}{U_{\mathrm{k}}}\right)^2\,dU. \tag{3} \]

\(n\) is the number of elementary sections over the length of sample \(e\);

\(\overline{U}\) is the mean value of the breakdown voltage of wire samples (from scratch tests);

\(\sigma\) is the root-mean-square value of the breakdown voltage;

\(\lambda\) is the parameter of the exponential law;

\[ \lambda=\frac{1}{E_{\mathrm{p}}\cdot X_{\mathrm{p}}}, \tag{4} \]

where

\(E_{\mathrm{p}}\) is the electric strength of the insulating gap between two conductors;

\(x_{\mathrm{p}}\) is the mean distance between conductors in the slot.

Expressions 1, 2, 3 make it possible to establish that the reliability of random-wound windings is influenced by the following principal factors:

1. The quality of the wires and of the impregnating compound \((\overline{U}, \sigma, E_{\mathrm{p}})\).
2. The slot fill factor and the number of conductors in the machine winding \((X_{\mathrm{p}}, N)\).
3. The level of commutation overvoltages and the number of individual coils in a phase \((U_k)\).
4. The level of winding-manufacturing technology \((a)\).

The purpose of the present work is to analyze those reserves for increasing reliability that can be used in production without changing the design of the machine.

A study of the scatter in the quality of the enamel wires supplied, and of fluctuations in the level of technology, showed that responsibility for the appearance of defects in machine windings lies both with the plant supplying the enamel wires and with the plant manufacturing the electric motors.

Figure 1 presents the dependence of the defectiveness of PETV— —f—35 wire of diameter 1.08 mm, tested after winding placement \((q_{\mathrm{out}})\), on its defectiveness determined in the as-supplied condition \((q_{\mathrm{in}})\). The tests were carried out according to the method described in [3]. In this case, defectiveness was taken to be the probability that a wire specimen 125 mm long would be punctured by a voltage \(U \leq 1\ \mathrm{kV}\).

\[ q=\frac{m}{n'}, \qquad \text{where} \tag{5} \]

\(m\) is the number of specimens punctured by a voltage \(U \leq 1\ \mathrm{kV}\);

\(n'\) is the number of specimens tested.

The dependence \(q_{\mathrm{out}}=f(q_{\mathrm{in}})\) was established on the basis of tests of 22 batches of wire, with a sample size of 200 specimens in each batch.

The existence of such a dependence is confirmed by an estimate of the correlation ratio \(\eta\):

\[ \eta = 0.756 \pm 0.06 \]

The first region of the curve \(q_{\mathrm{out}}=f(q_{\mathrm{in}})\) \((0<q_{\mathrm{in}}<0.1)\) indicates that the quality of the turn insulation of the windings depends entirely on the level of the technological process for manufacturing the winding and is practically independent of the quality of the incoming wire. At \(q_{\mathrm{in}}>0.1\) (the second region), the defectiveness of the incoming wire begins to affect the quality of the turn insulation of the windings.

An analysis of the quality of the enamel wire entering the plant showed that approximately 30–40% of it has \(q_{\mathrm{in}}>0.1\), which lowers the mean value of \(q_{\mathrm{out}}\) and, consequently, the reliability of the turn insulation \(R_{\mathrm{w}}\). In this connection, organization of incoming inspection of the wire should be considered expedient.

Preliminary calculations carried out using the data for these batches of wire as an example show that rejecting wire with \(q_{\mathrm{in}}>0.1\) will make it possible to reduce \(q_{\mathrm{out}}\) by 20% and to increase the reliability of the AO2—31/4 machine from 0.838 to 0.896. The rejected wire can be used in manufacturing the windings of a machine of another type, with more favorable design data (slot fill factor, a smaller number of conductors in a section, a larger number of sections in a phase, etc.), or for transformers, where layer winding excludes the appearance of pairs of conductors with a greater potential difference.

Another reserve for improving the reliability of the turn insulation of windings is to raise the level of inspection of the turn insulation [4].

Existing methods for checking the turn insulation of asynchronous-motor windings (SM-1, SM-2 apparatus) mainly involve recording metallic short circuits between turns, rather than detecting weak points in the turn insulation. At present, at the Sibelektromotor plant, a rotor-type installation proposed by us is being tested, intended for testing turn insulation with high voltage of increased frequency.

Fig. 1.

The method is based on the principle of inducing a high voltage in the stator winding.

A low voltage of increased frequency, 1000 Hz, is generated by a high-frequency generator and applied to the three-phase rotor winding inserted into the stator. In this case the rotor winding is connected in a star, and the stator winding is open. Thus, the circuit is a transformer with three secondary windings. If there are no defects in the insulation, the load in all phases of the rotor winding is symmetrical. The appearance of a turn-to-turn short circuit in one of the phases of the stator winding is equivalent to switching in an additional secondary winding of the transformer operating in short-circuit mode. This leads to asymmetry of the currents in the rotor circuit and makes it possible to record the turn-to-turn short circuit.

One of the tasks in raising the level of inspection of turn insulation is to establish the optimum test voltage. For this purpose we measured the dependence of turn-insulation rejects on the voltage in one phase of the winding.

Table 1 gives the test results. At each voltage, 60 stators were checked.

Table 1

The reject percentage practically begins to increase only at a voltage of 1700–1800 V. This is explained by the fact that the breakdown voltage of the air gap between two defective conductors has a certain minimum; at normal pressure it cannot be less than 400–450 V [5]. In our case, when the winding has 4 sections, the maximum possible voltage acting between two conductors in a section is equal to \(0.25 U_{\mathrm{f}}\).

The existence of the probability of ninefold overvoltages (the amplitude value of the phase voltage in this case is 2488 volts) also confirms the advisability of increasing the level of test-

…voltage for turn insulation. To compare the effectiveness of the existing and proposed methods for monitoring turn insulation,

Table 2

accelerated tests of machines for service life and reliability were carried out. The test results are given in Table 3.

Table 3

These results show that the detection of hidden defects during testing reduces the intensity of failures not only during the run-in period, but also during the period of physical wear of the winding insulation. Consequently, even minor defects not detected during testing gradually develop in operation and lead to failure more rapidly than defects formed as a result of insulation aging during operation of the machine.

Thus, defects in a section began to appear at voltages

\[ U_{\text{sec}} = 0.25\,U_{\phi} = 0.25 \times 1700 = 425\ \text{volts}, \]

which corresponds to theoretical concepts. At a voltage below 1700 V per phase, only metallic short circuits can be detected in the winding.

In the general case, increasing the test voltage leads to an increase in the fraction of rejects; however, the relative increase in rejects at voltages above 2500 V per phase decreases significantly. This is associated with the completion of detection of the bulk of the defects. Thus, the experimental data, as well as consideration of operating conditions, make it possible to assume that for random-wound windings the level of the test voltage per phase should be no less than $(0.4—0.45)n$ kV, where $n$ is the number of separate sections in a phase.

In order to establish the level of switching overvoltages, we carried out oscillographic studies of the amplitude and duration of overvoltages during switching on, switching off, and reversal of the motor in no-load mode. Table 2 summarizes the results of studies carried out on an AO2—31/4 motor. The probabilities of amplitudes and durations of overvoltages were determined by analogy with [5], with the number of oscillograms for each operation equal to 200.

Conclusions

1. A correlation has been established between the defectiveness of enamel wires in the as-delivered state and after winding placement.

2. It has been established that, in order to improve the reliability of random-wound windings of asynchronous electric motors, it is advisable to organize statistical incoming inspection of wire insulation in the as-delivered state.

3. The probabilities of the amplitudes and durations of switching overvoltages in the winding of the asynchronous electric motor AO2—31/4 have been determined.

4. A method is proposed for testing turn-to-turn insulation with an increased voltage.

References

1. Yu. P. Pokholkov, The influence of winding-and-insulation work and insulation quality on the reliability of windings of electrical machines. Dissertation, Tomsk, 1966.

2. A. S. Gitman, Yu. P. Pokholkov, E. K. Strelbitskii. A mathematical model of the reliability of turn-to-turn insulation of random-wound windings of electrical machines. Izv. TPI, vol. 172, 1967.

3. O. P. Muravlev, Yu. P. Pokholkov, E. K. Strelbitskii. The influence of winding-and-insulation work on the breakdown voltage of insulating materials. Elektromekhanika, No. 1, 1966.

4. Yu. P. Pokholkov, E. K. Strelbitskii. The influence of the level of testing of turn-to-turn insulation on the reliability of random-wound windings of asynchronous electric motors. Izv. TPI, vol. 172, 1967.

5. G. I. Skanavi. Physics of Dielectrics (Region of Strong Fields). Fizmatgiz, 1958.