Investigation of Defects in Turn and Ground Insulation of Random-Wound Windings of Induction Motors
Abstract Generated abstract
This paper examines defects in turn and ground insulation of random-wound stator windings in induction motors, with attention to failures during the run-in period. It proposes estimating insulation defectiveness from breakdown-voltage distribution laws rather than relying only on fixed-voltage breakdown counts, and applies Pearson curve fitting and chi-square checks to data from successive manufacturing stages of AO 32-4A motor windings. The results indicate that winding insertion causes the greatest damage to wire insulation, while use of softer tools and reduced slot fill can lower defectiveness for some insulation components. The method is presented as a way to reduce test scope and error when comparing materials, impregnation compounds, and manufacturing technologies.
Full Text
STUDY OF DEFECTS IN TURN AND GROUND INSULATION OF RANDOM-WOUND WINDINGS OF ASYNCHRONOUS ELECTRIC MOTORS
Yu. P. Pokholkov, E. K. Strelbitskii, A. S. Gitman, O. P. Muravlev
(Presented by the scientific seminar of the Departments of Electrical Machines and General Electrical Engineering)
Analysis of failures of asynchronous motors in service shows the presence of a clearly pronounced run-in period [1], with most failures occurring because of breakdown of the turn insulation.
The reliability of turn insulation during the run-in period can be estimated by means of the expression
\[ R = 1 - K + K(1 - mq^2)^N. \]
where \(q\) is the fraction of wire of length \(l\) having bare spots;
\(m\) is the probability that bare spots coincide in two contacting conductors;
\(N\) is the number of pairs of contacting wires of length \(l\);
\(K\) is the coefficient of detection of defects during the run-in period. It depends on the operating conditions of the motor and on the quality of the impregnating compound.
A similar formula can also be written for the reliability of ground insulation.
Usually \(q = 0.001 \ldots 0.20\) and can be estimated from the results of testing a single wire in pieces.
The quantity \(mq^2\) represents the defectiveness of a twist made of wires of length \(l\).
It is preferable to carry out tests on twists because of the instability of the coefficient \(m\). An experimental estimate of \(q\) at a fixed test voltage is given by the expression
\[ \hat{q} = \frac{n_q}{n}, \tag{2} \]
where \(n_q\) is the number of defective specimens,
\(n\) is the number of specimens tested.
The required amount of testing for determining \(q\) with a specified accuracy \(\varepsilon_q\) at confidence level \(\alpha\) is calculated by the formula
\[ n = \left( \frac{\operatorname{arg}\Phi_\alpha}{\varepsilon \sqrt{q'}} \right)^2, \tag{3} \]
where \(\operatorname{arg}\Phi_\alpha\) is the argument of the Laplace function for the value \(\alpha\);
\(q'\) is the actual value of the defectiveness.
The values of $n$ for $\alpha = 0.95$ and $\varepsilon = 0.1$ are given in Table 1.
Table 1
| $q'$ | 0.08 | 0.04 | 0.02 | 0.01 | 0.004 | 0.001 |
|---|---|---|---|---|---|---|
| $n$ | 99 | 141 | 197 | 281 | 364 | 884 |
For very small values of $q$, and all the more so of $mq^2$, the volume of tests, when oriented toward determining the number of breakdowns, becomes inadmissibly large.
In this connection it seems advisable to use information on the law of distribution of breakdown voltages and to determine $q$ and $mq^2$ by the formula
\[ q = \int_{u_1}^{u_2} f(u)\,du, \tag{4} \]
where $f(u)$ is the distribution function of the breakdown voltage;
$u_1$ is the lower limit of the breakdown-voltage distribution;
$u_2$ is the test voltage.
Formula (4) is written under the assumption that $f(u)$ is monotonic on the interval $(u_1, u_2)$. Consequently, the principal task in determining the defectiveness of windings is to establish the law of distribution of the breakdown voltage.
To study the defectiveness of insulating materials, the Tomsk Polytechnic Institute, together with the Special Design Bureau for Electrical Engineering at the “Sibelektromotor” plant, studied the technological process for manufacturing the stator windings of AO 32-4A motors at the following stages: 1) incoming inspection, 2) after the preparatory operation, 3) after laying the winding.
When removing insulation from the slots, measures were taken against additional damage to it. Testing of the materials was carried out in accordance with the existing standards.
On the basis of the results obtained, the statistics were calculated: $\bar{u}$—the mean value of the breakdown voltage, $\sigma$—the standard deviation, and the principal moments $r_3$ and $r_4$, which showed that the distribution of the breakdown voltage of the materials and of the wire insulation can be described by K. Pearson curves (Table 2).
Calculations of the curve parameters were performed on the “Promin” electronic digital computer. The agreement of the obtained distributions of breakdown voltage with the experiment was assessed by means of the $\chi^2$ criterion. The value $P(\chi^2) > 0.05$ for all curves, which makes it possible to regard the discrepancy between the theoretical and experimental curves as insignificant [2].
The obtained curve equations were used to calculate the fraction of defective insulation by formula (4), using numerical integration on the electronic digital computer.
The wire insulation is damaged most severely in the process of laying the winding in the stator slots. Obviously, this is caused by the metal tool used in laying.
To determine ways of increasing reliability, we made an attempt to reduce the defectiveness of the insulation by using a “soft” tool and by reducing the slot fill factor $K_z$ by 2.6%.
Table 2
Change in the parameters of the breakdown-voltage distribution of frame and turn insulation
| Distribution type and parameter | PEVА-2 wire — with varnish | PEVА-2 wire — after winding on a template | PEVА-2 wire — after laying the winding | Electrocardboard 0.1 — from a pack | Electrocardboard 0.1 — after laying the winding | Varnished cloth LSE 15, 0.15 — from a roll | Varnished cloth LSE 15, 0.15 — after forming coils | Varnished cloth LSE 15, 0.15 — after laying the winding | Electrocardboard 0.15 — from a roll | Electrocardboard 0.15 — after forming phase windings of coils | Electrocardboard 0.15 — after laying the winding |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Type of curve | IV | IV | IV | I | I | I | I | VII | IV | VII | |
| $\bar{U}_{\mathrm{br}}$ (kV) | 3,83 | 3,13 | 2,52 | 2,27 | 1,78 | 8,94 | 8,46 | 5,99 | 3,02 | 2,43 | 1,97 |
| $\sigma$ | 0,934 | 1,007 | 1,185 | 0,517 | 0,426 | 1,184 | 1,107 | 1,261 | 0,575 | 0,581 | 0,592 |
| $r_3$ | 0,661 | 0,102 | 0,493 | 0,623 | 0,588 | −0,586 | −0,507 | −1,293 | −0,078 | 0,061 | 0,295 |
| $r_4$ | 4,224 | 3,549 | 3,683 | 2,791 | 3,107 | 2,905 | 2,856 | 5,055 | 3,939 | 4,190 | 3,678 |
| $f_0$ | 0,0267 | 0,0262 | 0,0730 | 0,3728 | 0,4772 | 0,182 | 0,1755 | 0,2034 | 0,3799 | 0,3776 | 0,3621 |
| $l_1$ | 8,380 | 7,309 | 10,839 | 0,965 | 1,478 | 11,244 | 11,652 | 36,534 | 0,815 | 3,086 | 1,530 |
| $l_2$ | — | — | — | 4,744 | 5,877 | 2,931 | 3,693 | 1,279 | — | — | — |
| $a_1$ | 5,698 | 8,132 | 8,875 | 0,563 | 2,173 | 4,502 | 5,348 | 14,582 | 5,700 | 5,028 | 6,924 |
| $a_2$ | — | — | — | 2,768 | 8,672 | 1,173 | 1,695 | 0,510 | — | — | — |
| $P(\chi^2)$ | 0,49 | 0,95 | 0,6 | — | — | 0,1 | 0,5 | 0,1 | 0,1 | 0,1 | 0,2 |
The following were used as “soft” tools:
1) a rubber-coated hammer,
2) a fiber slot-insertion plate,
3) a steel reverse wedge coated with a thin layer of epoxy compound.
The values of \(q\), calculated from the experimental data, are given in Table 3 for the insulation after insertion of the winding.
Table 3
| Technology | PEVA-2 wire | electrocardboard 0.1 | varnished cloth LSEО, 15 | electrocardboard 0.15 |
|---|---|---|---|---|
| Rigid tool, normal \(K_3\) | 0,0182 | 0,00586 | 0,00307 | 0,0496 |
| Soft tool, normal \(K_3\) | 0,0035 | 0,00225 | 0,00234 | 0,0375 |
| Rigid tool, reduced \(K_3\) | 0,0075 | 0,0130 | 0,000 | 0,0250 |
| Soft tool, reduced \(K_3\) | 0,0009 | 0,0038 | 0,000 | 0,0170 |
The use of a soft tool substantially reduces the defectiveness of the wire insulation and has little effect on the main ground insulation—LSEО 0.15 varnished cloth and 0.15 electrocardboard. Reducing the slot-fill factor noticeably decreases the defectiveness of 0.15 electrocardboard, which is adjacent to the slot walls.
The described method for determining the fraction of defective insulation made it possible to substantially reduce the scope and error of tests when evaluating the effectiveness of the influence of various impregnating compositions on insulation defectiveness. Table 4 presents the results of tests of PEVA-2 wire twists removed from machines. The tests were carried out under normal conditions and at elevated humidity. The sample size was 250 specimens. The distribution \(u_{\text{pr}}\) is described by a Pearson type I curve.
Table 4
| Test conditions | not impregnated | varnish 447 | varnish 321 t | epoxy compound (ED-6, phthalic anhyd.) |
|---|---|---|---|---|
| Normal | \(0,132 \cdot 10^{-3}\) | 0 | 0 | 0 |
| After 48 h exposure under conditions of 100% humidity | \(0,584 \cdot 10^{-2}\) | \(0,95 \cdot 10^{-3}\) | \(0,70 \cdot 10^{-3}\) | 0 |
Conclusions
- The use of the breakdown-voltage distribution law makes it possible to substantially reduce the scope of testing when evaluating insulation defectiveness, as compared with tests at a fixed voltage.
-
The use of a “soft” tool and a reduction in the slot fill factor make it possible to outline ways of increasing the reliability of random-wound windings of electrical machines.
-
The defectiveness of insulation may be regarded as a criterion in studying the effectiveness of the application of new materials and technologies.
REFERENCES
- VNIIEM Report No. OAB 143, 673, 1963.
- A. K. Mitropolsky, Technique of Statistical Calculations. Fizmatgiz, 1961.