CH678988A5 - - Google Patents

Description

5

10

15

20

25

30

35

40

45

50

55

CH 678 988 A5

Description

The invention relates to improving the fire resistance of coated electrical conductors placed in hot service.

It applies in particular to dry power or distribution transformers, whose working temperatures conventionally exceed by more than a hundred degrees centigrade the ambient temperature (working temperature of the order of 140-150 ° C generally) . Also, and for convenience only, reference will be made for the rest of the description to the example of the aforementioned transformers.

In this type of transformer, usually reserved for a voltage range from 3 to 36 kV, the winding is embedded in an insulating synthetic resin with dielectric properties and having several millimeters of thickness (from 2 to 5 mm, for example). recalls that, in addition to its function of electrical insulation, the resin has a role of protection against humidity and dust which could lower the breakdown voltage. It also protects the electrical windings against an ambient environment polluted by aggressive chemical agents and plays an important role of mechanical resistance by ensuring the fixity of the turns between them in the winding.

However, we know that, under accidental extreme conditions of use, or following an anomaly, transformers can burn. The study of their fire behavior made it possible to realize that, when the coating resin burns, it often continues to burn, even if the fire which caused it has ceased, because, from a its own critical temperature, the resin becomes flammable in air.

At this stage, a few reminders concerning the constitution of these resins may prove useful for the following.

Current coatings for dry transformers are thermosetting resins, obtained by heating an initial mixture composed of a resin proper (generally an epoxide) and a hardener, such as anhydride. A classic example is provided by the diglycidyl ether derivatives of bis-phenol A, more commonly called DGEBA, and formed by reaction between bis-phenol A and hepichlorohydrin (see USP 3,202,947, for example). These resins are most often crosslinked and made insoluble and insoluble by the addition of amines and low molecular weight polysulphides (ARALDITE® resin, for example). Other examples are found among thermosetting polyesters. The weight proportions are conventionally 50% for the resin and 50% for the hardener.

Generally, loaded resins are used: the starting liquid resin receives, before the addition of the hardener, a filler, often mineral, for example quartz flour (silica), or glass wool, in weight proportions which are of the order of 3 charges per 1 of resin. In the initial mixture, for example, there will be 20% resin, 60% silica and 20% hardener.

The role of this filler is to improve the thermomechanical behavior and to absorb part of the heat of polymerization of the resin already, which avoids the formation of cracks. It also has a role of mechanical reinforcement, because the uncharged resin can remain of soft consistency at the operating temperature of the transformers.

On the other hand, rather elastomers are used, such as silicone resins, or polyester resins, which are thermoplastics, in the case in particular of coating cords or wires, in order to allow them a certain flexibility. Examples are given by E.P.R (Ethylene-propylene-rub-ber, rE.V.A. (Ethylene vinyl acetate), or crosslinked Polyethylene).

The invention does not differentiate according to the type of resin (thermosetting, thermoplastic or elastomer), ia insofar as this resin is an insulating material intended to be used "hot" and therefore therefore exhibits good thermosetting behavior. mechanical at the high working temperatures indicated above. Also, for convenience of language, we will indifferently call coating resin or "charged resin" the constituent material of the final insulating coating formed by the mixture: actual resin, reinforcing filler (if any), and possible hardener (usual additives included, such as accelerator, flexibilizer, etc.).

The object of the invention is to improve the fire behavior of coated electrical conductors, in particular the windings of dry transformers, by increasing, at their operating temperature, the thermal stability of the flame retardant coating resin used.

To this end, the subject of the invention is an electrical winding coated in a charged insulating resin containing a flame retardant substance by the formation of water when the temperature rises, such as hydrated alumina, characterized in that a fraction of the charge, equal to at least 20% of the total weight of the coating resin, consists of said flame retardant substance, which has been partially dehydrated beforehand in an amount such that it does not cause untimely formation of water in the resin as long as the coated winding is in service under its normal temperature conditions.

The invention also relates to a process for preparing an insulating resin charged with this type for the coating of electrical windings, according to claim 4.

The flame-retardant role of alumina trihydrate, as an adjuvant to coating resins for electrical cables or for transformers, is already known, for example from the document USP 3 202 947 already mentioned (and the teaching of which is incorporated here by reference). But, never before, to the knowledge

2

5

10

15

20

25

30

35

40

45

50

55

CH 678 988 AS

from the inventors, we have been able to highlight the advantage that we could paradoxically expect from better thermal stability of the resin by partially dehydrating such an adjuvant beforehand, while it acts as a flame retardant precisely by forming water.

The invention will be well understood, and other aspects and advantages will appear more clearly, in the light of the detailed description which follows, given with reference:

- Tables I, II and III presented in the body of the text and expressing the good resistance to fire of a coating resin according to the invention;

- in Table IV, also presented in the body of the description, and giving the main characteristics of the profile of the water departure with the rise in temperature for a certain number of flame retardants;

- in figs. 1 and 2 showing the evolution over time of the weight loss of various usable flame retardants, by formation and elimination of water, resulting from their "hot" instability when they are heated to constant temperature.

We successively take up the two essential characteristics of the invention set out above:

ADDING A SUFFICIENT QUANTITY OF A FLAME RETARDANT BY FORMING WATER DUE TO HOT INSTABILITY.

The flame retardant added may be hydrated alumina AI2O3, nHgO with n = 1.2, or 3 of ia magnesia bihydrate, zinc borate, or any other material known for its self-extinguishing properties by elimination of water and preferably also capable, like silica, of reinforcing the resin. Preference may be given to alumina trihydrate, which has proven to be the most effective flame retardant, and which, moreover, produces little or no smoke.

The water formation reaction can be written:

2 AI (OH) 3 - »AI2O3 + 3 H2O

The speed of this reaction increases in the direction of the arrow with the temperature, and its endotherm minimizes or even prevents the reaching of the ignition threshold of the resin.

We will see later that this speed does not evolve linearly with temperature, but that an intense starting peak of water is formed which is characteristic of the substance, and of which a decisive advantage is that it appears at precisely located temperatures in the overheating zone at risk, therefore below the critical inflammability temperature of the coating resin.

AI (OH) 3 can be easily mixed with the starting mineral filler, since they are both in the form of solid powder or fine grains.

Instead of 60% by weight of SÌO2 in the final resin, it is replaced, for the most part, with the flame retardant substance. This produces, for example, a mixture representing 50% of AI (OH) 3 and only 10% of S1O2. However, experience has shown that the amount of AI (OH) 3 can be reduced to 25% (therefore 35% of SÌO2), without significantly affecting the qualities of good resistance to fire of the coating. , due to the presence of the flame retardant.

These values, established for an initial load of 60% by weight, can of course be modified if this proportion varies.

Tables I, II and III below provide indications and quantified results of laboratory tests carried out on a dry transformer, showing the effect of flame retardants (here AI (OH) 3 and Zinc Borate) on the values parameters recognized as representative of the fire resistance of materials and in particular of the coating resins of the windings of dry transformers, namely the Oxygen Index, the Combustion Speed and the Higher Calorific Power.

3

5

10

15

20

25

30

35

40

45

50

55

60

RR

CH678 988 A5

Table 1:

Oxygen Indices (LO.)

Measuring temperature

20 ° C

80 ° C •

150 ° C

200 ° C.

Indica d! 02 (I.O.) minimum imposed

30

27

23.5

21

Alumina Tri-Hydrated.

Total charge rate 70%

70% of S1O2

-30

-28

-24

-19

20% of SiÓ2 and 50% of AI (OH) 3

-39

-37

-32

-31

Total charge rate 60%

60% of S1O2

. -26

/

l

/

50% of Si02 and 10% of Ai (OH) 3

-25

/

/

/

40% Si02 and 20% AI (OH) 3

-27

/

/

l

35% of Si02 and 25% of Al (0H) 3

-30

-28

-25

-20

25% SÌO2 and 35% Al (OH) 3

28-31

28-31

24-27

19-22

20% Si02 and 40% AI (OH) 3

30-33

29-33

25-28

23-24

10% Si02 and 50% AI (OH) 3

32-35

30-32

27-29

24-26 '

Zinc Borate

Total charge rate 65%

10% of Si02 and. 55% Borate.

-32.5

no values

Total charge rate 60%

10% SiQ2 and 50% Borate

-30.3

no values

60% Borate

-33.6

no values

The coating material was prepared in the following manner: the mineral filler (here silica), after having been mixed with the flame retardant in an adequate amount, was kneaded, for half, with the liquid resin (epoxy resin marketed by the Swiss company CIBA-GEIGY under the name "ARALDITE CY 225") and, for the other half, with the hardener, also in the liquid state (anhydride marketed by the aforementioned company under the name "DURCISSEUR HY 227") . The two mixtures were then combined and the whole was kneaded, then put in the oven (80 to 150 ° C) for solidification.

For these tests, the measurement methods complied with UTE NF T51-071 standards at 20 ° C and

EDF HN20 M40 at 80,150 and 200 ° C.

Rq-

- The ranges of values reflect the fact that the I.O. measured in this case depended on the origin of the commercial alumina that was used.

- The boxes marked with a “/” reflect measures deemed unnecessary. The corresponding values, taking into account that obtained at 20 ° C, are immediately unacceptable, because too much below the t.O. minimum imposed.

It is immediately known that for a total charge of 60% the minimum content of AI (OH) 3 to be respected is 20-25%. Below the minimum values imposed on the O2 Index are no longer effectively insured. Additional measurements, not recorded here, have made it possible to show that, with an overall load of 70% by weight (1st part of the table), the threshold value of AI (OH) 3 drops to 20%. We also observe that, as far as Zinc Borate is concerned, the minimum threshold is much higher: 50% by weight at least, which translates, as we knew, to a greater efficiency of Alumina trihydrate.

The Combustion Speeds recorded in Table II below were measured in the apparatus used for the determination of Oxygen Indices on specimens in the form of elongated platelets, 100 mm long, 6.5 mm wide and 4 mm thick. The test pieces have two marks following one another along the length, the first being located 10 mm from one end, the second at 60 mm. The combustion time of the test piece between the two marks is noted at room temperature and the average combustion speed (in mm / s) is deduced therefrom as a function of the rate of oxygen in an oxidizing oxygen mixture and nitrogen.

4

5

10

15

20

25

30

35

40

45

50

55

CH678 988 A5

Table II:

Combustion speeds.

02 rate

35%

40%

45%

50%

60%

Maximum Authorized Combustion Speed (in mm / s)

0.15

0.30

0.45

0.6

0.9

Tri-Hydrated Alumina

Total charge rate 70%

70% SÌO2

0.28

0.37

0.47

0.58

1.05

20% Si02 + 50% AI (OH) 3

0.06

0.10

0.13

0.19

0.25

Total charge rate .60%

60% of Si02

-0.35

- 0.50

- 0.60

-0.75

-1.0

50% SiCbf tÖ% AI (OH) 3

-0.35

-0.45

-0.60

-0.75

-1.0

40% sìo2 + 20% AI (OH) 3

-0.20

"0.35

-0.45

-0.55

-1.0.

35% SÌO2 + 25% AI (OH) 3

-0.18

-0.28

-0.38

-0.49

-0.95.

•

25% SÌO2 + 35% AI (ÓH) 3

0.14-0.17

0.20-0.23

0.27-0.30

0.34-0.37.

0.56-0.59 • '

20% of sìo2 + 40% of AI (OH) a •

0.13-0.16

0.19-0.20

0.24-0.27

0.30-0.35

0.48-0 £ 6

10% deSI02 + 50% AI (OH) 3

0.07-0.10

0.14-0.16

0.20-. 0.23

0.26-0.29

0.37-0.40

Zinc Borate

Total charge rate 65%

10% of Si02 + 55% of Borate

-0.16

-0.21

-.0.30

-0.8

Total charge rate 60%

10% of Si02 + 50% of Borate

~ 0.25

-0.43

-0.65

-0.98

60% Borate

-0.25

-0.31

- 0.52

-0.96

Rq. the indications of pairs of values have the same meaning as in the previous table.

As can easily be seen, the values given in Table II corroborate those in Table I by showing that, with regard to the criterion “Burning speed” also, the “floor” value to be respected for the quantity of AI ( OH) 3 in the starting mixture is substantially 25% by weight (slightly less than 50% for Borate).

These conclusions remain entirely valid in light of Table III below, recording the results of the series of tests carried out on the 3rd parameter retained, the Higher Calorific Power (P.C.S.). As will be seen, the maximum allowable value of 11 MJ per kg of material is never exceeded.

Table III:

Higher Calorific Power (P.C.S.)

(Test method: adiabatic calorimetry according to the UTE NF M 03-005 standard.)

Total charge rate: 70%

Total charge rate: 60%

Tri-hydrated alumina

70% sìq2 20% SÌQ2 +

60% so2

40% sìo2 +

35% SiÔ2 +

25% SiQ2 +

10% SiQ2 +

50% Al

10% Al.

25% Al ..

35% ai:

50% Al ..

-8 -7

-11

-11

-11

-11

-11

Zinc borate

10% SÌO2 + 55% Bor.

10% SÌO2 + 50% Borate

60% Borate

-10

-11

-11

It is therefore clear to see clearly on these tables the influence on the fire resistance of the addition to the starting resin, in adequate proportions, of a flame retardant material by formation and elimination of water, in accordance with the invention.

5

5

10

15

20

25

30

35

40

45

50

55

60

65

CH678 988 A5

These results reflect a strong release of water from the coating resin, when the latter reaches abnormally high temperatures. Such a phenomenon acts as a self-controlled heat absorber, which delays and slows down the combustion of the charged resin, and gives it the desired self-extinguishing character.

Of course, an abundant formation of water molecules within the mass itself is not without consequences for the quality of the coating resin. This degrades during such a process and, generally, can no longer be reused for the future. The transformer must then be replaced or reconditioned.

The loss in weight of the flame retardant during the rise in temperature is a good indicator of its capacity to form water. It will be noted, as the specifications of the suppliers show, that, in the case of Al (OH) 3, the weight loss is already close to 30% at 300 ° C. It also takes place practically only at this temperature in the form of a narrow peak and of large amplitude, which testifies to the liveliness and intensity of the phenomenon, when this level of temperature, characteristic of the agent flame retardant used, is reached.

This is true regardless of the variety of commercial alumina trihydrate used, as shown more precisely in Table IV below.

This table of values is given for information only, based on the indications of the suppliers of PAI (0H) 3 sold under the name "ALCOA" according to varieties, the commercial references of which are given in the columns of the table to identify them. .

Table IV:

Analysis of endcrthermal peaks of water formation

(The temperature rises were carried out from 25 to 600 ° C under a constant progression of ° 10 ° C / min.)

.MIS

S65 / 4Q

C31

M6

M15S

S65 / 15Ö

MEDIUM SODA

Temp. start of peak (° C)

196

205

216

180

195

188

197

Temperature of the room (° C) '

372

385

353

382

355

370

353

Temp. max pio (° C)

316

312

314

314

311

309

312

AH of the peak. (in kJ / g)

1.03

1.01

1.07

1.06

1.03

1.0

1.02

For the effects expected by the type of flame retardant recommended by the invention to be fully satisfactory, it is not only necessary that the intense starting peak of water develops at adequate temperatures, that is to say between the point of normal hot operation and that where the coating can ignite, but still that a formation of water is not really carried out except in the event of abnormal overheating. In other words, it is necessary to avoid any risk of premature degradation of your coating resin which could be caused by untimely formation of water from ambient temperature to that of normal hot operation of the transformer.

It is this difficulty which is resolved by the second main characteristic of the invention, namely, moderate pre-aging of the resin, as will be seen in more detail below:

2) PRE-PARTIAL DEHYDRATATIQM OF THE QUANTITY OF ADDED FIREPROOF SUBSTANCE

This is to cause a prior dehydration to prevent it from occurring later in the transformer. However, this dehydration should only be partial, since it is also sought at high temperature, when the transformer heats up abnormally and there is a risk of ignition.

This dehydration can advantageously take place by prior heating of the AI (OH) 3. The objective to be achieved is, not a total elimination of the water likely to form in a range of temperatures from amber to (a "hot" operating temperature of the transformer (and which will be called pictorial way “volatile water” to mark the fact that it must leave at low temperature), but a sufficient elimination so that the residual “volatile” water is present in too small quantity to lead to a degradation of the resin. could observe, in fact, that, when the alumina trihydrate had not been preheated before being added to the mineral filler, bursts of the resin coating your electrical windings of test transformers intervened at operating temperature, which forced them to be scrapped.

A convenient way to achieve partial dehydration of the flame retardant by preheating is to consider its weight loss curve over time. For a substance used for the first time, it is advantageous to proceed in two stages:

6

5

10

15

20

25

30

35

40

45

50

55

CH678 988 A5

- A first, on an analysis sample, intended to determine the amount of water which is eliminated during an extended stay at constant temperature, which is that (or close to that) of normal hot operation of the transformer;

- A second step, this time processing all of the material, consisting of heating it to a relatively high temperature in order to quickly reach, therefore under industrial conditions, a weight loss corresponding to the water elimination value determined in the previous phase.

Of course, this second step will be repeated with each resin preparation, while the first is only really necessary once to characterize a type of flame retardant that has never been used before.

The curves of fig. 1 and 2 illustrate this first phase of study on samples by showing the trend, at certain temperature values, of the evolution of weight loss as a function of time for certain temperature values. The curves are set on some of the varieties of Alumina trihydrate "ALCOA" in Table IV identified by their commercial references at the right end of each curve. Three temperature values were considered: 140, 160 and 180 ° C in order to properly cover the usual "hot" operating conditions of the transformers. To avoid unnecessary overloading, the three corresponding families of curves have been separated in the two figures: FIG. 1 groups families at 140 ° C (broken lines) and 160 ° C (solid line); the family at 180 ° C. appears alone in solid lines in FIG. 2.

As we can see, all these curves are advantageously of general logarithmic appearance with a very rapid growth at the beginning, followed by a level slightly inclined on the horizontal, this level being all the sooner reached when the working temperature is high. Thus at 180 ° C (fig. 2), most of the “volatile” water (around 80%) is already formed after only 140 h, over the more than 700 h of total duration of the tests. The weight loss then ranges between 5.3% and 2.5% approximately, depending on the type of alumina.

The existence and the stability of the bearings are verified by testing aluminas previously heated to a higher temperature. Tests were thus carried out on two samples of the variety "Ml 5" maintained for 18 h, one at 180 ° C the other at 200 ° C. The results appear in fig. 1 in the form of two straight lines (A) and (B) having ordinates of origin of 1.6% and 4.2% of weight loss for 180 and 200 ° C respectively. One notes their quasi-horizontality, which reflects the insensitivity of the samples to a second heating at a lower temperature, due to the fact that almost all of their “volatile water” at 140 ° or 160 ° C has actually was eliminated during the first heating.

We also observe in fig. 2, that at the end of 18 h at 180 ° C., a sample “M15” is barely halfway up its rapid growth curve for weight loss by elimination of water.

It will be immediately understood that these curves have a favorable characteristic appearance which makes it possible to deduce very easily the degree of prior partial dehydration which must be achieved. We can thus, for example, choose as a criterion, the start of the plateau and retain as the proportion of “volatile” water to be eliminated, the value given by the ordinate of this start of the plateau.

Thus, the figures show that for the variety "S65 / 40" we can retain 2.5% weight loss, for a transformer operating temperature of 140 ° C, 4% at 160 ° C and 5.5% at 180 ° C.

Likewise, the variety "M15" will accommodate a prior weight loss of 2% for operation of the transformer at 160 ° C and 3.5% at 180 ° C.

At 140 ° C, the curve of this variety has a less typical appearance. It should be noted, however, that the value of 0.7% obtained after approximately 500 h may be perfectly suitable.

To fix the ideas, it can therefore be said overall that the weight loss prior to targeting ranges between 0.5% and 10% approximately for all the flame retardants selected for the implementation of the invention.

The transition then to the industrial phase consists simply in dehydrating the mass of flame retardant substance accordingly by rapid heating techniques in the oven, with monitoring of the weight loss, for example by gravimetry using an automatic balance, the tray is placed in the oven enclosure.

For information, in the case of "M15" type aluminas, the weight loss of 3.5%, sought for operating the transformer at 180 ° C, could be achieved after 6 hours of heating only at 200 ° C.

This heating operation will of course be all the more rapid the higher the heating temperature. However, for obvious reasons of preservation of a high potential for water formation, necessary in the event of abnormal overheating, care will be taken not to exceed much, and preferably to remain below, the temperature at the start of the peak d intense removal of water characteristic of the flame retardant used and some values of which are given in table IV above.

If one does not wish to have to operate by gravimetry, in particular because of too large quantities of material which could possibly be to be treated, it will be advantageous to proceed by an intermediate stage making it possible to translate the targeted weight loss into a heating period . This can be done using a second measurement sample of the flame retardant concerned, of which the weight proportion of water to be removed is known and which is subjected to rapid heating at a determined high temperature. This heating will operate with a continuous weighing of the sample in order to be able to

7

5

10

15

20

25

30

35

40

45

50

55

CH 678 988 A5

measure the time required to achieve a weight loss corresponding to the known proportion of water to be eliminated. The value of the measurement defines the duration of the heating operation, at a temperature strictly identical to that of the above intermediate operation, to which the mass of flame retardant substance to be treated will be subjected.

If necessary, it will be possible to ensure the proper execution of the operation by determining by gravimetry at high temperature, for example 12 ° C., the remaining water content of a sample taken for this purpose from the mass of treated flame retardant . By comparison with the initial total water content (usually of the order of about 20 to 35% by weight) which has previously been determined analogously on a reference sample, it is deduced therefrom that the quantity of water " volatile ”effectively eliminated is in line with that intended.

It should be noted that the heating time is not entirely independent of the particle size of the material. It has been observed, during the tests, that a coarse particle size causes a greater weight loss than a fine particle size for a given heating time.

It will also be observed that the weight loss values determined above, by reading the test curves made from samples, in no way represent an upper limit not to be exceeded. Only, as these values correspond to the start of the plateau, there is in principle no use in continuing the heating for a long time to gain a few tenths of a percent, which, in any case, are probably too insignificant to cause a water departure. capable of adversely affecting the quality of the coating resin.

It will certainly be understood that these curves, valid at normal operating temperatures of the hot transformer, reflect the behavior of so-called "volatile" water, at these temperatures. At higher temperatures, the levels are higher and are reached more quickly, in particular at the temperature of the peak characteristic of the flame retardant used and which, as we have seen, is in the vicinity of 30 ° C. for all the varieties of AI (OH) 3 studied.

The flame retardant thus pretreated becomes ready for use. It remains to complete the preparation of the coating resin in the usual way: the mineral filler, after having been intimately mixed with an adequate amount of partially pre-dehydrated flame retardant, is divided into two bright parts. One is then introduced into the pure resin bath, and the other into the hardener, also in the liquid state. The two mixtures are kneaded separately to form two solid-liquid suspensions, then combined into a single mixture which is kneaded in turn to ensure good homogeneity. The resulting pulp is then poured into a mold, in which the electrical winding of the transformer that is to be coated has been previously placed. After casting the mold is placed in the oven for solidification of the resin. After diversion and cooling, the resin block incorporating the winding is demolded and can then be mounted in the transformer intended to receive it.

It goes without saying that the invention cannot be limited to the examples described above, but extends to numerous variants or equivalents, in so far as the characteristics set out in the claims which follow are respected.

In particular, the invention is not limited to the transformers themselves. It should indeed be understood, by this designation used here, more broadly all of the inductive electric apparatus or equipment able to work at relatively high temperatures, from 100 to 200 ° C. as we have seen, and whose electrical windings can be drowned in a block of insulating resin.

Similarly, although the invention was initially conceived for the applications of thermosetting-cissable resins (coating of inductive windings, in transformers in particular), it in fact relates to all dielectric coating materials. It is of particular interest in the case of electrical installations having nominal long-term operation at high or medium temperature. In other words, it will be particularly advantageous to use it for coating resins which already have good thermomechanical properties.

Claims (11)

Claims 1. Electric winding coated in a charged insulating resin containing a flame retardant substance by formation of water when the temperature rises, characterized in that a fraction of the charge, at least equal to 20% of the total weight of the resin coating, consists of said partially dehydrated flame retardant substance in an amount such that it does not cause untimely formation of water in the coating resin which may degrade the quality thereof, as long as the winding is used in its normal temperature conditions. 2. Coiled electrical winding according to claim 1, characterized in that said flame-retardant substance has, compared to its non-dehydrated state, a weight loss of about 0.5 to 10%. 3. Coiled electrical winding according to claim 1 or according to claim 2, characterized in that the flame retardant is alumina trihydrate. 4. A method of preparing an insulating resin charged with coating electrical windings with a view to obtaining a coated electrical winding according to claim 1, according to which a flame retardant substance is added to the initial resin, in addition to the filler. formation of water when the temperature rises, process characterized in that said substance is added at a rate of at least 20% of the total weight of the coating resin, and in that previously said said dehydrated substance in quantity 8 5 10 15 20 25 30 35 40 45 50 55 60 65 CH678 988A5 such that there is no untimely formation of water which could be detrimental to the behavior of the coating resin, as long as the electric winding remains in service at its nominal operating temperature. 5. Method according to claim 4, characterized in that the flame retardant is subjected to a partial partial dehydration treatment until it has a weight loss of between 0.5 and 10% of its initial weight. 6. Method according to claim 4 or 5, characterized in that one uses, as flame retardant, alumina trihydrate. 7. Method according to claim 6, characterized in that one determines the degree of preliminary partial dehydration of the flame retardant which is used for the first time, by taking a measurement sample which is subjected to heating prolonged at constant temperature, lower than that of the peak of elimination of clean water from the flame retardant used, and by considering the pace of the evolution of the weight loss of said sample over time. 8. Method according to claim 4 or 7, characterized in that one carries out the partial partial dehydration of the flame retardant by subjecting it to rapid heating until its loss in weight corresponds to the amount of water to eliminate according to a proportion previously determined on a measurement sample, the evolution of the weight loss as a function of which has been followed over time, during prolonged heating at a temperature below that of the peak of elimination of clean water from flame retardant used. 9. Method according to claim 5 or 8, characterized in that one carries out the partial partial dehydration of the flame retardant by subjecting it to rapid heating until its loss in weight corresponds to the amount of water to eliminate according to a proportion previously determined on a measurement sample whose evolution of the weight loss as a function of time has been followed, during prolonged heating at a temperature below that of the peak of elimination of water suitable for flame retardant used. 10. Method according to ia claim 9, characterized in that said rapid heating is carried out for a period which has been predetermined by a heating operation carried out at an identical temperature and carried out with weighing of a sample of the flame retardant substance used up to 'that it has a weight loss corresponding to the amount of water that is to be eliminated. 11. Method according to claim 10, characterized in that said quantity of water to be eliminated has been determined beforehand by monitoring the development over time of the weight loss of a measurement sample which is subjected to prolonged heating at constant temperature, lower than that of the peak of elimination of water specific to the flame retardant used. 9