TRANSFORMER RATINGS AND TRANSFORMER LIFE

Edward Simonson. The National Grid Companv DIC.

Introduction

This paper explores the relationship between transformer ratings and life and
highlights some of the issues involved in determining transformer loading based on
the standard IEC 354 equations. On-line and off-line transformer ratings methods
are described. Finally, a “wish list” for the future is presented from a transformer
user’s perspective.

The basis of transformer ratinqs

Any electrically conducting material which passes a current (with the possible
exception of superconductors) results in the generation of heat. If the heat can be
removed at the same rate as that at which it is produced, then an equilibrium
conductor temperature is reached which is governed by the various thermal
conductivities in the system.

It is the aim of the designer to ensure that such temperatures within a transformer
are kept to reasonable levels when loaded up to continuous maximum rating. In
practice, a power transformer has a significant thermal capacity largely arising from
the insulating oil. This allows for the possibility of loading for short periods beyond
the continuous rated nameplate value. Any requirements for such loading requested
by the transformer operator also need to be taken into account as part of the
transformer design in order to ensure that defined temperature limits are not
exceeded.

Limitations on acceptable conductor and oil temperatures will typically be stated in

the transformer specification, and these in turn are likely to be based on
recommendations in such documents as the international standard loading guide for
oil immersed power transformers, IEC 354‘’’.

The effects of temperature in a transformer

Physical and chemical effects governing the interaction between materials are
generally temperature dependent and chemical reaction rates typically increase with
increasing temperature.

It is usually accepted that the ageing of the paper insulation system is such that the
stated transformer life (whatever that may be) can be achieved for a continuous
maximum hotspot temperature of 98°C. Beyond this, it is assumed that the rate of
ageing doubles for every increase of 6°C. At temperatures of the order of 150°C,
accelerated ageing tests in the laboratory demonstrate that the useful life of the
paper may only be a few days. The mechanical strength of the paper arises from
long chain cellulose polymers. The chain length can be characterised by a degree of
polymerisation (DP) measurement that gives values of 1000 or more for new paper.
The cellulose structure of the paper degrades at high temperatures until at a DP of
about 200, the paper has no remaining mechanical strength. This clearly limits the
life of the transformer and is one of the governing factors on the maximum load that
should be used.

Paper ageing can be considered to be cumulative but there are other temperature
related factors which may only be transient, and therefore not have a lasting effect

711
on transformer life, provided that they are not sufficiently severe to cause the
immediate demise of the transformer. One such factor is the production of gas in the
oil. Significant gassing can start to occur at temperatures in the region of 140°C to
160°C and above. This can seriously compromise the dielectric strength of the
insulation system and lead to premature failure. However, gas bubbles produced in
the windings will tend to occur in the hottest regions nearer the top and may quickly
be drawn off into the cooling system. Likewise, bubbles resulting from high
temperatures caused by stray loss heating of the core or other metal parts need not
pass through areas of high electrical stress at all and may thus offer a lower risk of
failure. Of course, unless the gas is removed in some way or re-dissolves before
circulating out of the coolers, there is scope for serious problems when it re-enters
the transformer at the bottom of the windings. The rate of gas production could thus
be as important as the total amount produced, but either way, although such
considerations provide another limit on transformer ratings, they need not have any
direct effect on transformer life.

The other major transformer component to be considered is the insulating oil. Oil is
subject to degradation due to direct thermal effects, and enhanced oil temperatures
are likely to accelerate other ageing processes. The condition of the oil can affect
the ability of a transformer to withstand emergency overloading. The concentrations
of moisture, gases and impurities due to ageing are all important. However, unlike
the paper insulation, the oil can be reclaimed or changed where this is deemed
beneficial. Oil condition can have an effect on loading, although this is not an area
where conclusions can easily be drawn.

Additional factors to be taken into consideration are possible limitations imposed by

bushings or tap changers. These will be rated to meet the required cyclic load
capability of the transformer and any other short period loads required by the
operator. However, bushings have a much lower thermal capacity than that of the
transformer and therefore there may be little scope for extrapolating to loads beyond
those stated in the specification. Tap changers are typically tested up to twice rated
current when on a fixed tap, but are likely to be restricted to less than this for tapping
operations. This can impose a complicated restriction on calculation of loading guide
tables where tap changing is inhibited above certain loads.

Ratinqs calculations based on IEC 354

Short period and cyclic loadings are typically calculated in accord with the principles
of IEC 354. The loading guide tables therein are based solely on the equations for
winding hotspot and oil temperatures, without regard for the other factors discussed
above. This is because the effects such as heating by stray losses are very
dependent on design and in any case not easy to calculate. However, in
constructing loading guide tables, the transformer user needs to be aware of
restrictions on loading other than winding temperature and the circumstances in
which those restrictions might apply.

The biggest uncertainties in relation to winding temperatures are likely to be due to

the estimation of hotspot factors and corrections due to variations in temperature of
load losses and oil viscosity. Although the IEC guide makes recommendations for
how to approach these latter issues, it is also suggested that more detailed studies
may in individual cases may lead to a better approach. As for the estimation of
hotspot factors, fibre optic measurements in transformer windings performed by
NGCf2)in co-operation with manufacturers suggest that there may be quite some
variability in the actual values to be used. However, those same measurements
have always indicated that the hotspot temperature (if indeed the fibre sensor is in

712
 the correct location) is below the desired maximum temperature when corrected to
standard conditions. This provides a significant comfort factor at least as far as
loading up to nameplate values is concerned.

Loading up to nameplate load tends to be taken for granted irrespective of

transformer age and it is the short period loadings which are of particular interest.
This is especially important from the viewpoint of those planning and operating the
system, where a transformer may need to carry a load in excess of the nameplate
value for some while following a fault, before post fault action can be taken to restore
the load flows to lower values. The question therefore arises as to the point to which
the IEC equations can be sensibly extrapolated. The loading tables in the IEC guide
adopt an upper limit of 2 pu load, but stray loss heating which is not included in the
loading guide equations might impose restrictions below this value.

For transformers fitted with thermal protection devices such as winding temperature
indicators (WTls),the immediate practical limitation is that the transformer should not
trip when a short-term emergency load is applied. Hence it is important to know top
oil time constants under natural and forced cooling with some degree of accuracy,
and also to model the behaviour of the top oil temperature with change in cooling
state in a more sophisticated fashion than in IEC 354. Fibre optic temperature
measurements in transformers are helping to increase understanding of these
issues. It should however be borne in mind that in using the WTI as a protective
device the implicit assumption is being made that the WTI is providing an accurate
estimate of winding hotspot temperature.

On-line and off-line ratinas methods

NGC’s circuit thermal monitor (CTM) presently monitors 60 transformer and cable
circuits. It uses real time values of load and ambient temperature to calculate on-line
ratings. This provides the advantage over off-line methods that actual conditions
prevailing on the day can be used. It also provides the ability to assess the physical
health of a transformer where excessive loading might be considered detrimental. It
is possible to monitor Hydran values for some of the transformers on the CTM so
that a continuing record can be built up of the behaviour of the transformer with load.

NGC is presently replacing the transformer loading code TRALCOI with a new
. version for calculating off-line ratings. The new versian, TRALC2, allows modelling
of loaded secondary and tertiary windings and includes a discrete tap model. The
thermal equations are those in the IEC guide: however it is possible in future to add
other thermal solvers to allow for improvements in the IEC modelling. Cyclic ratings
can be calculated for any 24 hour profile of ambient temperature, secondary (and
tertiary) load and voltage on the HV winding. It is also possible to calculate spot
ratings (i.e. the maximum sustainable load for a specified time until a defined limit
temperature is reached) based on any input load curve. A matrix of such values can
also be determined based on defined pre-loads, ambient temperatures and post-fault
load durations - in effect a loading guide table. The electrical model makes
allowance for the operation of WTI and overcurrent protection and includes inhibition
of tap changing above given loads. A simple model is included to calculate the
effects of stray loss heating. All input and output values are stored in a Microsoft
Access database for retrieval at any time.

The pu winding currents relative to nominal rated values are only the same for a two
winding transformer without taps. Thus the cases for loaded tertiary and tapped
windings are of particular interest. In these situations it is necessary to consider
each winding individually, rather than adopting a given load as applying to the
transformer as a whole, as the IEC loading guide tables tend to imply.

TRALC2 is necessarily dependent on measured data from type test certificates and it
is intended that TRALC2 will allow off-line transformer ratings to be determined as
precisely as this information will allow, whilst including all relevant system operating
conditions.

A “wish list” for the future

It is not possible to cover in detail everything of relevance to transformer ratings and

life management in the space of a few pages. However, the transformer operator is
often required to work plant harder and this requires the ability to “work plant
smarter” i.e. sufficient understanding of the issues is required to use the loading
capability of a transformer to the full whilst appreciating the effect that this may have
on asset life. Thus it is pertinent to enquire what might be useful routes to follow to
improve existing knowledge.

There are three main areas to consider:

Firstly, the actual temperatures achieved under all loading conditions need to be
known with a good degree of accuracy, and this includes both winding hotspot
temperatures and temperatures of any other metal parts caused by stray loss
heating. This requires finite element modelling to determine stray losses and
calculation of heat generation and oil flow within the windings using programs such
as fEFLOW2‘’’. The results of the modelling should be supported and refined by
direct measurement with fibre optic sensors.

Secondly, appropriate values of the constants to be used in the IEC equations need
to be determined. The oil and winding exponents may be broadly independent of the
exact design details for transformers of specific types, but hotspot factors are highly
dependent on design and the best values to use for each case should be
established. A further step is to demonstrate by comparison with factory and site
measurements that the IEC equations are providing a good representation of the
actual thermal behaviour.

Thirdly, the assumption that the WTI provides a good simulation of the IEC hotspot
temperature calculation needs to be examined. Even if it does, there is often a
significant discrepancy between the cooler top oil, which is typically used in winding
gradient calculations, and the tank top oil on which the WTI value is based in
practice.

It is desirable to obtain more fibre optic temperature measurements within

transformers for reasons given above, and ideally in one or more of each design, to
assess the degree of variability (or otherwise) introduced by manufacturing
tolerances.

In the future, it may be possible to build large power transformers with distributed
fibre sensors in the windings and this would provide the opportunity to understand
how the temperatures throughout windings depend on load and cooling state. It
would also enable the location of hotspots to be demonstrated: at present point
sensors are placed in windings in the anticipated hotspot location but this might not
be known with certainty.

714
Ultimately, temperatures from fibre sensors could be fed directly into on-line ratings
software, and artificial intelligence techniques could be used to characterise the
thermal performance of the transformer based on past history. This would enable
more accurate ratings predictions to be made in the future.

At present, it would be possible to obtain an improved determination of the thermal

behaviour in factory tests than is usually done. For off-line modelling purposes, it
would be helpful to have measurements of winding gradients at the extreme tap
positions and at mid tap for both natural and forced cooled conditions. This would
help to determine the effect of the tapping winding (if any) on the results.

Winding time constants are implicitly determined at the end of heat run tests in order
to extrapolate winding resistances back to the point when the load is removed.
These time constants and measurement of oil time constants should be made known
to the transformer user so that the short period thermal response to load can be
adequately assessed.

A future generation of microprocessor controlled WTls could incorporate winding

gradient and time constant values, or even read fibre optic measurements of winding
temperatures directly, whilst giving cumulative assessments of use of life based on
integrating the temperature against time relationship.

Further work should be done to determine the range of applicability of the IEC
equations. The most likely way that this could be achieved would be by experiment
on a transformer at the end of its service life and which had been retro-fitted with
fibre optic probes. A variety of loads up to and beyond 1 pu could be applied and the
thermal response determined.

This paper has chiefly addressed the issue of the effect of rating on transformer life.
Of course it is possible to stand the whole question on its head and ask what the
effect might be of transformer life on the rating. This is probably the more
controversial question, particularly since the assumption is usually made that
transformers are always good up to nameplate load, if not beyond. There is a variety
of techniques available to assess the state of a transformer: oil analysis can reveal
much about the state of the oil and the paper, whilst Recovery Voltage
Measurements give information about the moisture content in the paper. Tests on
an aged transformer could be used to obtain useful information on the ability of such
a unit to withstand short-term overloads as well as test the applicability of IEC
equations.

Conclusions

This paper has explored some of the issues surrounding transformer life and ratings.
This is an involved area that requires good understanding of the thermal behaviour of
the transformer, both in terms of winding heating and stray loss heating of other
metal parts, and the effect that temperature has on the various ageing processes.

The international loading guide IEC 354 provides guidance on cyclic and short term
loading. The equations therein are necessarily simplistic but provide a basis for
establishing a transformer rating. However, improvements in the IEC calculations
can be obtained by detailed studies of each transformer family.

Ratings can be calculated by on-line or off-line methods. On-line methods offer the
advantage that real time ambient temperatures and load can be used to obtain
ratings appropriate to actual operating conditions. Off-line methods allow the user to
explore different modelling possibilities such as enhancements to the IEC equations.

Future improvements in ratings can be obtained by better characterisation of

transformer thermal performance through computer modelling and fibre optic
measurement, and by additional factory measurements during heat run tests. When
combined with an understanding of the ageing processes of transformer
components, and the effects that ageing may have on the ability of the transformer to
carry loads, an optimised use of the loading capability and transformer life may be
obtained.

Acknowledaements

The work was carried out at National Grid Engineering and Technology and the
paper is published with the permission of The National Grid Company plc.

References

(1) IEC 354, Loading Guide for Oil-immersed Power Transformers, 1991.

(2) Thermal Capability Assessment for Transformers, E A Simonson and J A

Lapworth, Second International Conference on the Reliability of Transmission
and Distribution Equipment, IEE Conference Publication Number 406, pp
103-108, March 1995.
0 1998 The Institution of Electrical Engineers.
Printed and published by the IEE, Savoy Place, London WC2R OBL, UK.