412 ABSTRACTS OF PAPERS

MECHANICAL STRESSES IN TRANSFORMER WINDINGS

By E. BILLIG, Dr.Tech., Member.*
(ABSTRACT of a Transmission Section paper which was published in June, 1946, in Part II of the Journal.)
ARRANGEMENT OF WINDINGS the windings in such a way as to prevent an undue amount of
Although power transformers have attained a high degree of unbalance in any tapping position, though this migjrt "preclude
reliability in service, breakdowns nevertheless do occur from time perfect balance for any one position.
to time under abnormal conditions. Such conditions arise when
transformers are subjected to frequent short-circuits at their ELECTROMAGNETIC FORCES AND STRESSES
secondary terminals, with the supply voltage maintained at Mapping out the space surrounding the conductors in lines of
nearly its full value. force, Maxwell has shown that the magnetic forces may be
The magnetic energy of a transformer is stored in the iron core visualized as—
and in the magnetic field in the space surrounding the windings.
The forces involved, therefore, greatly depend on the relative (a) A tendency for thefluxlines to contract like elastic bands.
disposition of the several parts. "Concentric" winding arrange- (b) A tendency for adjacent flux lines to repel each other.
ments with circular coils are now almost universally used for The former tendency causes a pressure to be exerted on any
large transformers, this being the shape any coil tends in general individual winding when carrying current, whilst the latter pro-
to assume under the action of the electromagnetic forces. duces a pressure perpendicular to the flux lines at any point.
Some of the gravest dangers arise from windings which are Flux lines passing between two conductors carrying opposing
not properly "balanced" magnetically, that is, where primary currents will thus produce forces pushing the conductors apart.
and secondary windings are not arranged symmetrically, or Table 1 shows the magnitude of this radial force between two
where local flux concentrations occur due to the preponderance simple concentric windings for various sizes of typical 3-phase
of the ampere-turns of one winding over the other. Such dis- 50-c/s transformers when their secondary terminals are short-
symmetry always produces additional stray flux, eddy-current circuited.
loss, local heating and increase in mechanical stress. TABLE 1
Fig. 1 shows some of the usual tapping arrangements. The
axial forces acting on the windings vary with the tapping position. Three-phase kVA 100 1000 10000 100000

Reactance, % 3-75 5-5 8 10

Effective width of duct, inches.. 1 2 3 6
Crest radial force, tons 44-6 152 697 2 783

This value is the theoretical maximum which, under practical

conditions, will never be reached owing to resistance losses and
to the voltage drop at the primary terminals of the transformer.
(a) (b) (c) Excessive currents can also be caused by internal faults which
may occur from one of several causes—
(1) Breakdown of the insulation between turns, coils or windings,
or from any live part to the frame or to the iron core (earth).
(2) Faulty operation of the tap-changing gear.
(3) Accidental short-circuit between tapping leads or across the
terminals to which these are connected.
=«-4b
Owing to the concentration of heavy currents within a small
space, the forces and stresses produced by an internal fault—
especially across a few turns only—are usually much more
severe than in the cases previously considered and are likely to
result in permanent deformation, decrease of clearances and
further breakdowns. Once started, internal faults thus have a
tendency to spread progressively unless the supply is quickly
interrupted.
All the forces described above act directly on the conductors
and must be overcome by the mechanical strength of the con-
ductors themselves and of the supporting structures between
turns, coils, windings and the core. These supports consist of
insulating material and represent the weakest link mechanically.
As the slightest dissymmetry in the windings gives rise to forces
which increase rapidly with the displacement, the whole structure
should be so designed as to avoid any movement whatsoever,
Fig. 1.—Tapping arrangements. even under extreme conditions. Mechanically weak coils, such
as special tapping coils, should never be placed near the main
Except for special winding arrangements, such as shown at (h) duct, where the leakage field is strongest, but on the outside of
and (k), it is impossible to obtain perfect balance between the the outer winding, where they are subjected only to compara-
windings in all tapping positions. It is thus advisable to arrange tively little stress.
* British Electrical and Allied Industries Research Association. Insulating supports between coils warrant careful considera-
 ABSTRACTS OF PAPERS 413
tion: they must not be too far apart, as this would result in an
excessive bending stress in the conductors; and their bearing Tie rods
area must be large enough to prevent any excessive compression
of their own or of the conductor insulation. On the other hand,
sufficient clearance between the supports must be allowed so as
to avoid interfering with the cooling of the coils. Clamping
arrangement.
SHRINKAGE AND CLAMPING OF WINDINGS
The large pressures produced during a fault may cause some
permanent set in the insulating material. Conductors which
have moved slightly under the action of the electromagnetic
forces whilst the fault lasted, would return to their original
positions only if all the deformations produced were entirely
elastic. Due to permanent set in the insulation, however, some
portions of the windings will be left in a slightly shifted position.
Quite apart from any thermal or other ageing effects, there is-
thus a certain amount of shrinkage produced by every short-
circuit. Excessive currents, too, may appear on energizing a
transformer. Although considerably below the short-circuit
peak, they may amount to several times the rated current. As
there is no relief from opposing ampere-turns in any other
winding, the contractive effects will be relatively large. Inte-
grated over the whole life of the transformer, the shrinkage pro-
duced by frequent switching may be considerable, and may lead
to serious trouble when a short-circuit occurs.
The amount of shrinkage will be different for the various coils
according to the amount and nature of the insulating materials
used in their construction. Under the same pressure the h.v.
winding will shrink more than the l.v. winding.* Under the
influence of their own weight or of clamping springs, the coils
then drop by a corresponding amount, so causing an axial dis-
placement of the two windings, the very condition which must
be avoided, as it results in increased forces, further shrinkage Fig. 2.—Clamping arrangement.
and greater displacement on the occurrence of the next short-
circuit. The vicious circle so started is likely to cause damage transmitting the pressure within the coil, the stress/strain
to the conductor insulation by continuous abrasion and to lead characteristic of the former will be much steeper [see Fig. 2(c)].
ultimately to complete breakdown. Coil and tie-rod will come to rest both with the same length
To avoid trouble from shrinkage every care must be exercised, h0, i.e. at a point where their characteristics intersect. The
not only to balance the forces by proper design, but in the con- pressure Po in the coil is then equal to the tension in the tie-rod.
struction of the windings to use only suitable insulating materials After releasing the pressure, the tie-rod would contract to a
and in such a way that shrinkage in service is reduced to a length hs and the coil expand to length hc—if there were no
minimum. A case where particular difficulties are likely to be permanent set in either.
met concerns the support of windings with a small radial depth Now assume that an electromagnetic force Ph is generated
such as cylindrical tapping coils, especially if located in a strong within the coil by heavy unidirectional current tending to con-
magnetic field. The practice of using end rings cut from tract it. Coil and spring will contract slightly and come to rest
Bakelized-paper cylinders or similar material is not to be recom- at a point I. The compression of the coil is very slightly in-
mended, as they are likely to deteriorate in time under the com- creased, and the tension in the rod, owing to its steep charac-
bined influence of hot oil and mechanical pressure applied in an teristic, greatly decreased, the difference between the two cor-
axial direction, i.e. edgewise to the paper laminations. Materials responding to the internal contraction Ph. If the force Ph—
for the supports should be pre-shrunk for a prolonged period owing, for instance, to some turns being cut out from the
until no more settling will occur under any conditions that can centre of the coil—is reversed, resulting in an outward thrust
arise during operation. on the clamping gear, the coil and the rod will slightly expand,
The purpose of a well-designed clamping structure is to prevent giving largely increased tension in the rod but again hardly
any movement within the windings. To investigate the behaviour affecting the pressure in the coil.
of the clamping gear when electromagnetic forces arise within It is clear that, owing to the "soft" characteristic of the insu-
the windings, assume the simplest case of a coil C clamped by lating material within the coil, the pressure there hardly varies
means of a tie-rod S [Fig. 2(a)]. Being subjected to mechanical and almost the whole of the electromagnetic force is met by a
stress, they will both show some elastic strain, as illustrated in change in the tension of the tie-rod. This holds good as long
the simplified diagram in Fig. 2(b), where the clamping gear is as the electromagnetic force Ph tending to contract the coil
indicated as a spring S. On assembly, the coil stack has an remains below the original clamping pressure PQ. Otherwise the
initial length hc. When pressure is slowly applied by tightening tension in the rod will disappear completely and the whole force
down, the stack is reduced in length. At the same time, the tie- Po will be thrust on the coU itself, as shown in dotted lines in
rod is subjected to tension. Owing to the vastly different Fig. 2(c). A sudden large increase in coil pressure would result
elastic behaviour of the metallic tie-rod and the insulating material in considerable permanent set. Assuming that to be the case, it
• Fortunately,1 this effect is largely counteracted by the proximity of the iron core, is clear that when the contracting force Pc has disappeared, both
which tends to decrease appreciably the pressure in the outer—h.v.—winding. coil and tie-rod will settle down to a slightly shorter length, with
414 ABSTRACTS OF PAPERS
a reduction of the clamping pressure to P'Q. Should that process further movement by the insulating spacers. The compressive
be repeated several times the clamping pressure inight be lost stresses are therefore considerably increased by the impact
completely, leaving the winding slack and vulnerable to subse- nature of the forces.
quent damage. A similar result might be expected from a slow Records should be kept of the performance of transformers
"creep" of the tie-rod. under short-circuit. Such records should include frequency
An immediate effect of the loss in clamping pressure would and severity of faults in the system, and state of windings as
be a certain slackness in the windings. At the occurrence of found by occasional inspection, such as development of slackness
the next fault an electromagnetic force of larger magnitude will and signs of movement in the windings. Quantitative results
again tend to compress the winding. Owing to the slackness could readily be obtained by fitting a simple strain indicator to
between adjacent coils, however, these will start closing up and the clamping gear. Useful information could be gathered from
gather considerable momentum before being restrained from a periodic check of the clamping pressure.

THE POWER RATING (THERMAL) OF RADIO-FREQUENCY CABLES

By SQUADRON-LEADER R. C. MILDNER, R.A.F.V.R., Associate Member.
(ABSTRACT of a Radio Section paper which was published in July, 1946, in Part III of the Journal.)
Coaxial or balanced cables are now widely used for coupling field. For Service use a standard ambient temperature of 55° C
transmitting equipments of medium and high power to their has been assumed, and power ratings have been worked out for
antenna systems. Such transmitters are used for broadcast, a limited number offlexiblecables which have been widely used
television, and radar for Service and civil use. High-frequency in the past few years. The cable is assumed to be working into
generators for induction or dielectric heating use cables to trans- a matched load and to have attained steady-state conditions:
mit power from the equipment to the point of application. In all the latter assumption is made because the thermal capacity of
of these applications the cables used may be of semi-air-spaced this type of cable is normally quite low. The ratings are shown
or of solid-dielectric types. graphically for a wide range of frequency for the cables when
Commercial production of these cables has reached the stage installed in air, and also, in a few cases where the construction
where the electrical characteristics at all operational frequencies is suitable, when laid in the ground. "Rating factors," to be
may be specified to reasonable tolerances, and this has permitted applied to the standard ratings, are proposed for dealing with
the determination of the maximum power which may safely be cases where the ambient temperature differs from the assumed
carried by such cables. This safe power-carrying capacity is value.
normally determined by the permissible temperature rise gener- Whilst a theoretical basis has been adopted throughout the
ated within the cable by the electrical losses of transmission. In paper in determining the attenuation and power rating, there is
exceptional cases, such as certain pulse transmitters of low duty- ample evidence available to confirm substantially the assump-
cycle, a limiting factor may be the maximum voltage which the tions made. The procedure adopted is necessary because the
insulation can safely withstand; this limitation is not considered precise distribution of the electrical losses in the cable has to be
in the present paper. assumed in arriving at the final temperature rise.
Temperature limitations may be set by the softening point of The effect of standing waves on the transmission line, due to
the thermoplastic materials used, by the necessity of avoiding the load not being matched to the characteristic impedance of
chemical or physical deterioration of the insulation or protective the cable, is briefly considered. This effect cannot be expressed
coverings, or by differential expansion of the component parts of in a concise form covering all the conditions which may be met
the cable. For Service use, a maximum temperature of 85° C is in practice, but the effects of two limiting cases are given in
proposed for solid-dielectric (polythene) cables. graphical form. It is recommended that, unless a more rigorous
Formulae for estimating the attenuation of various types of study can be made, the rating should be reduced by a factor
cable are given, and empirical factors for the resistance of equal to the standing-wave ratio.
stranded and braided conductors are proposed. Some con- Finally, the effect of longitudinal transfer of heat along the
sideration is also given to the behaviour, over a wide range cable is considered. It is shown that two factors are involved:
of frequency, of bimetallic conductors, such as tinned the exponential decrement of the power transmitted due to the
copper. The electrical and thermal constants of the various electrical losses in the cable, and the enhanced dissipation of
materials used in the construction of radio-frequency cables are heat which may occur at the cable terminations. These cor-
stated. rections are found to be unimportant under most conditions of
The methods used to determine the mean power rating of these operation, and no allowance need be made except for high-
cables are based upon principles established in the power cable attenuation cables.