OPTIMIZATION OF IMPREGNATION AND DRYING MODES FOR RANDOM-WOUND WINDINGS

I. A. VOLKOMIRSKII, N. M. NOVOSELOVA, A. Ya. TSIRULNIK

(Presented by the scientific seminar of the Department of Electrical Machines)

The reliability of winding insulation during machine operation depends substantially on the degree of solidity of random-wound windings. Solidity is achieved by impregnating the windings in electrical-insulating varnishes with subsequent drying, and depends both on the cementing properties of the varnish and on the impregnation and drying modes. To increase the reliability of windings, it is necessary to determine the optimal modes of the technological process of impregnation and drying of windings that provide a high degree of winding solidity.

This article presents the results of an experimental study of the possibility of improving the quality of windings when they are impregnated with ML-92 varnish (winding wire—PETV) according to the technology used at the Elektromashina plant (Prokopyevsk). According to this technology, the armature of a direct-current machine, heated to a temperature of 60–80°C, is immersed in cold varnish and held for 30 minutes, after which it is removed and the varnish drains off for 10–20 minutes. Drying is carried out in two stages. First, at a temperature of 60–80°C for two hours—to remove the solvent. Then at a temperature of 140°C for 14 hours—to carry out the process of varnish polymerization and its transition to a thermoset state.

The determination of the optimal conditions for the course of the process, under which the best quality is obtained, is carried out by conducting and analyzing the results of a multifactor experiment planned in accordance with the mathematical theory of experiment. The following parameters may be used as input parameters characterizing the technological process:

\[ \begin{aligned} \eta &\text{— viscosity of the varnish at } 20^\circ\text{C, sec according to VZ-4;}\\ t_{\mathrm{i}} &\text{— temperature of the article when it is immersed in the varnish, }^\circ\text{C;}\\ T_{\mathrm{p}} &\text{— duration of the impregnation process, min;}\\ t_{\mathrm{c}} &\text{— temperature in the oven during drying in the process of removing volatiles, }^\circ\text{C;}\\ T_{\mathrm{c}} &\text{— drying duration, h;}\\ t_{\mathrm{ц}} &\text{— temperature in the oven during the process of polymerization (cementation) of the varnish, }^\circ\text{C;}\\ T_{\mathrm{ц}} &\text{— duration of the cementation process, h.} \end{aligned} \]

For physical reasons, factors 2, 3, 4, and 5 were excluded from the number varied at the first stage of the study. It is undesirable to raise the temperature of the article when it is immersed in the varnish because of the increased intensity of evaporation of the solvent from the varnish. The duration of the process

impregnation can be considered sufficiently justified by existing production experience. In addition, this duration is small compared with the duration of the entire process. The levels of factors 4 and 5 are also sufficiently justified by production experience. An increase in the drying temperature leads to rapid formation of a surface varnish film, which makes it difficult to remove solvent from the internal volumes of the winding. The drying duration corresponds to the given drying temperature. Reducing the number of variables substantially simplifies the experimental design. The variable input parameters of the process are \(\eta = x_1\), \(t_{\mathrm{c}} = x_2\), and \(T_{\mathrm{c}} = x_3\).

It is expedient to carry out the investigation not on actual products, but on mock-ups. This facilitates evaluation of the result obtained, i.e., the degree of solidity of the winding. Solidity can be assessed by the magnitude of the forces required to pull out conductors placed between the conductors of the mock-up winding. Such a method of assessment follows from the procedure for evaluating the cementing ability of varnish according to GOST 13526-68. The mock-up was made as follows. The stator core of a size-4 asynchronous motor, preliminarily boiled along the outer surface with longitudinal seams, was cut into segments with a disk milling cutter. Each segment had 5 slots, into which PETV conductors of diameter \(1.6\) mm were laid with a slot fill factor of 0.75. The conductors protruded from the slot by 10 mm on one side and by 50 mm on the opposite side. In each slot, between the main conductors, five test conductors per slot were placed, penetrating into the slot over a length of 50 mm. After impregnation and drying of the mock-up, these conductors were pulled out on a tensile-testing machine to evaluate the degree of cementation under different impregnation and drying regimes. The bundles of conductors protruding from the slot were bandaged in order to imitate approximately the same conditions for varnish penetration into the slot as in the armature of a machine. The levels of variation of the variables are shown in Table 1.

Table 1

The experimental design is shown in Table 2 (experiments No. 1–8). Each experiment corresponds to one mock-up. After impregnation and drying of the mock-ups and determination of the conductor pull-out forces for each mock-up (experiment), the mean value \(Q\), the standard deviation \(\sigma\), and the coefficient of variation \(v\) of the conductor pull-out forces were calculated. These parameters are given in Table 2.

The distribution of pull-out forces in each experiment is shown in Fig. 1. The mean value of the force does not fully characterize the distributions. In some cases, with a larger mean, there may be a greater dispersion due to the appearance of small pull-out forces. This indicates a high probability that weak spots will occur in the winding despite good cementation on average.

A desirable distribution of forces is one characterized by a high mean with low dispersion. The coefficient of variation \(v\) decreases as the mean increases and the dispersion decreases, that

Table 2

Fig. 1. Distribution of the forces for pulling conductors out of the slot in each experiment

that is, it has good resolving power in evaluating the output quality in each experiment. This parameter was therefore adopted as the output parameter.

Regression analysis of the experimental results made it possible to obtain the dependence of the parameter $v$ on the input parameters in the following form:

$$ v = 0.1793 - 0.0265 \cdot x_1 - 0.1338 \cdot x_2 - 0.0748 \cdot x_3 + 0.0182 \cdot x_1x_2 - $$

$$ {} - 0.0067 \cdot x_1x_3 + 0.0767 \cdot x_2x_3 . \tag{1} $$

The change in the parameter $v$ when the input parameters change can be calculated from (2), which follows from (1).

$$ \Delta v = 10^{-4} \cdot [7 \cdot \Delta x_1 + 174 \cdot \Delta x_2 + 56 \cdot \Delta x_3 + 3.3 \cdot \Delta (x_1x_2) + $$

$$ {} + 0.46 \cdot \Delta (x_1x_3) + 59 \cdot \Delta (x_2x_3)] . \tag{2} $$

The factor $x_1$, when varied within the limits of the variation interval, does not exert a significant influence on the output parameter; therefore, in equation (1) it can be fixed at one level, for example, at the central level ($x_1=\eta=18$ sec). Then we obtain the following dependence:

$$ v = 0.1793 - 0.1338 \cdot x_2 - 0.0748 \cdot x_3 + 0.0767 \cdot x_2x_3 . \tag{3} $$

The isoquants of the surface described by expression (3) are shown in Fig. 2. The surface has a saddle-like shape with axes AB

Fig. 2. Isoquants of the response surface

and CD. To obtain better quality (a smaller value of \(v\)), it is necessary to move along the AB axis from point O toward point B without limit. However, function (3) describes the process adequately only within limits not exceeding the intervals of variation by a factor of 1.5–2. Motion along the AB axis toward point B means increasing the drying temperature while reducing the duration. At a certain elevated temperature the process changes qualitatively, and function (3) becomes inadequate. This does not permit unlimited movement along the AB axis with the expectation of obtaining better quality. For PETV wire, let us limit the temperature to \(160^\circ\text{C}\) and carry out experiments at three values of drying duration: 6 hours, 10 hours, and 12 hours. The results are shown in Table 2 (experiments 9, 10, 11) and in Fig. 1. The best result is obtained under the condition \(t_{\text{d}} = 160^\circ\text{C}\) and \(T_{\text{d}} = 10\) hours. Thus, by raising the drying temperature at the stage of varnish polymerization to \(160^\circ\text{C}\), we can reduce the required duration of the drying regime from 14 to 10 hours.

To verify whether drying at a temperature of \(160^\circ\text{C}\) for 10 hours has a negative effect on the turn insulation of the wire, a comparative experiment was conducted in which one batch of conductors, after impregnation, was dried at a temperature of \(140^\circ\text{C}\) for 14 hours, and the other batch at a temperature of \(160^\circ\text{C}\) for 10 hours. Both batches of conductors were cut from PETV wire from the same coil. After drying, the conductors were tested for breakdown and the distributions of breakdown voltages of the turn insulation of both batches of wire were compared. In both cases, no defective spots or microcracks in the insulation (low breakdown voltages) were found. This indicates that the drying regime at \(160^\circ\text{C}\) for 10 hours is permissible, which shortens the duration of the technological cycle for impregnating and drying windings of low-power machines by 4 hours.