You are on page 1of 11

EUROPEAN TRANSACTIONS ON ELECTRICAL POWER

Euro. Trans. Electr. Power 2010; 20:927–937


Published online 3 August 2009 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/etep.375

Method for the reduction of in-service instrument transformer explosions

Miroslav Poljak1*,y and Boris Bojanić2


1
Končar-Electrical Engineering Institute, Fallerovo šetalište 22, 10000 Zagreb, Croatia
2
Končar-Instrument Transformers, Josipa, Mokrovicá 10, Zagreb, Croatia

SUMMARY
The methodology and test procedure for the reduction of possible in-service explosions of instrument transformers to negligibly low values is
described. As the introduction, an interpretation of statistical results of a survey about different types of failure in instrument transformers is given,
showing that about 90% of them develop gradually, taking from several hours to several months. The International Electrotechnical Commission
(IEC) is about to adopt a standardised procedure requiring that instrument transformers should withstand the effect of an internal arc. The authors
are taking a critical view of the proposal; explaining its main flaws. On the basis of calculations and experimental research on models and 123 kV
current transformer it is shown that the mechanism causing explosions of instrument transformers cannot be simulated well enough in the way
provided for in the draft of the new IEC standard. A new concept for the reduction of explosions of instrument transformers in service called
‘explosion safe’ is proposed. It includes additional routine and special tests, and usage of devices for the detection or measurement of oil pressure
rise in transformers. The proposed tests are based on the investigation ‘Volt–Time Relationships for PD Inception in Oil–Paper Insulation’.
Copyright # 2009 John Wiley & Sons, Ltd.

key words: instrument transformer; failure; explosion; test; internal arc; Weibull distribution

1. INTRODUCTION

Electric system reliability is an important component in the requirements for electricity supply systems, which are becoming
increasingly complex. Failure of electric system components reduce directly its reliability, so that in recent years special attention
has been paid to their analysis, interpretation and creation of conditions for their prevention. Instrument transformers in the system
have an important role in revenue metering and protection of other system components, and that is why the study and statistical
analysis of incidents related to them always engage the attention of scientists and experts. Explosion of an instrument transformer is
the most difficult form of fault, and as a rule its consequences are:
- Direct cost of instrument transformer replacement.
- Direct cost of replacement of neighbouring equipment damaged in the explosion.

It is not rare that this type of fault also causes indirect costs through undelivered electricity during the elimination of its
consequences. There is also the possibility of injuries of personnel present in the station, although the probability for this is very low.
Thorough analysis of two international questionnaires organised in CIGRE, and several years of discussions within International
Electrotechnical Commission (IEC) working groups, have prompted the authors to make an additional theoretical and experimental
investigation into the phenomenon of instrument transformer in-service explosions. The most important results of the investigation
are systematised in a proposal of test procedures and other measures whose application could drastically reduce or even nearly
prevent explosions of instrument transformers.

*Correspondence to: Miroslav Poljak, Končar-Electrical Engineering Institute, Fallerovo šetalište 22, 10000 Zagreb, Croatia.
y
E-mail: mpoljak@koncar-institut.hr

Copyright # 2009 John Wiley & Sons, Ltd.


928 M. POLJAK AND B. BOJANIĆ

2. INSTRUMENT TRANSFORMER FAILURES—INTERNATIONAL SURVEY

The mentioned CIGRE survey covered failures of high-voltage (Um  72.5 kV) instrument transformers in about 15 countries. The
first survey covered 136 033 transformers in the period from 1970 to 1986 [1], and the other 131 207 transformers in the period from
1985 to 1995. Since the concepts of these two surveys were very different, it is difficult to compare their results. Therefore, the
authors will mostly refer to more recent investigations on the basis of statistical analysis of only major failures, their causes, methods
of detection and actions taken for their reduction.
A major failure is defined as a sudden explosive event that has caused an immediate emergency system outage or trip. Other
categories include minor failure and defects.
When speaking about statistical analysis of failures, we distinguish between a relative failure rate and a failure rate.
Relative failure rate is the ratio of the number of failures in the observed period to total number of transformers at the end of the
period, expressed in percentage. If the relative failure rate is divided by the number of years in the observed period, the result will be
the failure rate, which indicates the probability of failure occurrence in 1 year.
Table I shows the major failure rate (explosions) of instrument transformers in reference to both total recorded failures and
installed instrument transformers. The analysis is giver for different types of instrument transformers.
On the basis of figures in Table I it can be concluded that the number of explosions of instrument transformers in relation to the
total number of detected failures is 15.3%. The number of explosions is considerably higher with current transformers than with
other types, although the total failure rate is the lowest with current transformers. This can be explained by the fact that defects in
voltage and combined transformers are often manifested with secondary voltage changes, which enables an early response e.g. in the
form of open delta voltage measurement. It must be pointed out that the failure rate, which indicates the annual number of faults, is
in this concrete case 10 times less. So, on the basis of the data in Table I it can be stated that, statistically, in 1 year, 22.9 of 10 000
installed transformers will have a fault, of which 3.5 will explode.
The analysis of possible failure causes is very interesting and can be divided into seven categories:
(1) design faults due to errors in dielectric or mechanical calculations or use of inadequate materials;
(2) inadequate quality control including assembly errors, impurities in the insulating system, leakage of impregnating agents and
the like;
(3) aging if transformers that are older than 25 years;
(4) lightning impulse overvoltages;
(5) abnormal service conditions such as ferroresonance, cable or conductor breakdown, seismic events, faults in neighbouring
equipment and the like;
(6) inadequate maintenance including improper erection and commissioning;
(7) unknown causes.
Table II shows all the causes of major failure (explosions) in percentage in relation to total of failure in all types of instrument
transformers.
It is evident that about half of explosions of instrument transformers are caused by design faults. The cause of a large percentage
of explosions is unknown, especially with current transformers.
In inductive voltage and combined instrument transformers, lightning is a much more frequent cause of failure than for other types
of instrument transformers. The reason is probably the non-uniform distribution of voltages along the primary winding of the

Table I. Analysis of major failure (explosions) versus total numbers of failure and installed transformers.
Transformer Major Total Installed Major/total Major relative Total relative
type failure (1) failure (2) transformers (3) failure (1)/(2)  100 failure rate failure rate
(1)/(3)  100 (2)/(3)  100
Current 261 1024 65 190 25.5 0.40 1.57
Inductive voltage 85 537 19 215 15.8 0.44 2.79
Capacitive voltage 60 624 22 784 9.6 0.26 2.74
Combined 54 819 24 018 6.6 0.22 3.41
Total 460 3004 131 207 15.3 0.35 2.29

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
REDUCTION OF IN-SERVICE INSTRUMENT TRANSFORMER EXPLOSIONS 929

Table II. Causes of instrument transformer failure in %.


Primary cause Current Total nos Inductive voltage total Capacitive voltage total Combined total nos
of failure of failures ¼ 261 nos of failures ¼ 85 nos of failures ¼ 60 of failures ¼ 54
Design fault 34.5 54.1 43.3 59.3
Inadequate quality control 17.2 7.1 16.7 0.0
Ageing 10.3 1.2 26.7 0.0
Lightning 8.4 17.6 3.3 29.6
Operation outside specification 3.1 13.0 3.3 9.2
Inadequate maintenance 0.4 1.2 0.0 0.0
Unknown 26.1 5.9 6.7 1.8

inductive voltage transformers. The same applies to service conditions outside of the specification ratings, as in the case of
ferroresonance in inductive voltage transformers.
To take adequate measures for the prevention of instrument transformer failures, it is necessary to know, apart from the cause, also
how they were detected. An additional CIGRE survey shows quantitatively the ways in which faults are detected:
- 63% by regular visual inspections and maintenance;
- 13% by system protection devices;
- 11% by transformer monitoring (voltage transformers);
- 8% by some other method;
- 4% through fault in similar equipment;
- 1% through unplanned inspection of the station.
Taking into account the causes and ways of failure detection, it can be concluded that a large majority of failures slowly develops
over time. To this group belong all failures that were not detected by relay protection, and even a part of those are detected in that
way. The failures due to overvoltage and ferroresonance do not belong into the group of slowly developing ones. On the basis of
these facts, it can be concluded that 85–95% of instrument transformer explosions are the result of slow processes. Their slow
development enables their timely detection or prediction. As a rule, in transformers with paper–oil insulation, slowly developing
failures have, as a consequence, an increase of insulating liquid pressure. A pressure gauge or pressure switch [2] provides a simple
and relatively cheap means for the prevention of explosions.
Users’ reaction to failures and explosions can be also seen from the additional survey [1]. Most of them have improved the
maintenance and applied more efficient diagnostic methods. Some have changed the transformer ratings or even the type and the
manufacturer of their transformer. At the same time, requirements for manufacturer’s quality assurance have become more
stringent, and more weight is placed upon routine and type tests.

3. WHAT HAS BEEN DONE WITHIN IEC

In the beginning of 2000, an IEC document proposal was issued to include in instrument transformer standards the definitions,
requirements and tests related to explosions due to internal arcs. The initial proposal applied only to SF6-insulated transformers;
relied on practices in the United States and Canada, and the tests were, by their nature, special tests.
Two categories of protection were defined: internal arc fault protection class I and class II. The first category covers the
transformers in which internal arc can cause fracture of insulators and the enclosure, but the fragments must remain within a
specified radius, while for the second category no fracture is allowed. The current that initiates the arc equals the short-time short-
circuit withstand current (Ith) of 0.2 second duration, and the dynamic current equals 2.5 times the thermal current. The arc inception
shall be located in the highest dielectric stress area. The proposal had the following objectives:
(1) to investigate what the users want at that particular moment;
(2) to provoke a discussion about this issue with the result of producing a better proposal for a test;
(3) to raise awareness about these problems rather than define the degree of transformer safety.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
930 M. POLJAK AND B. BOJANIĆ

By a majority of votes, the members of IEC TC38 MT 30 concluded that requirements on internal arc should be extended to
transformers with oil–paper insulation. In 2001 a document very similar to the first draft was proposed, however with a following
difference: the categories ‘internal arc fault protection class I and class II’ were renamed to ‘explosion resistant’ and ‘explosion
proof’. This proposal tries to define the choice of place of arc inception, and for current transformers that place is identical to the
highest dielectric stressed area.
Since 1987 Électricité de France (EDF) has tested current transformers with Um  245 kV for internal arc withstand and in 1993
the testing has been expanded to current, voltage and combined transformers with Um  72.5 kV. The intent was to define
transformers with reinforced security. In contrast to the IEC proposal, EDF has tested transformers with short-circuit currents
statistically determined by taking into account several unfavourable conditions on the French network. The current is symmetrical
and the arc duration was from 0.525 to 2.5 seconds which is equal to the time necessary for the protection devices to clear the fault.
From 1985 to 1987 the tests were made with a current equal to Ith with an asymmetrical peak value of 2.5 times the symmetrical
component. The explosions during the test were violent and manufacturers had great difficulties to find an economically viable
solution. The analysis of tests made in EDF [3] shows that on the basis of tests on 20 transformers the pressure relief device will
operate in 15–100 ms, and that oil will catch fire in 45–125 ms, which leads to the conclusion that 0.2 second is adequately
representative for the testing.
It is further demonstrated that the energy generated during testing done under the IEC approach exceeds the one generated in
testing done under the EDF approach. It is also pointed out that with inverse-type current transformers, which exploded in the French
network, the fault almost always occurred at the top of insulating system that is outside the porcelain insulator. That was the reason
why that place was chosen as the area of arc inception. In voltage transformers with a closed core that place is in the tank at the
transformer bottom.
During drafting of the standard the most discussed issues were the place and the current of arc inception. The documents were
submitted three times to the IEC national committees for their remarks and discussion.
In the beginning of 2004, in one of the variants, it was proposed that for inverse-type current transformers the location of arc
inception should be the transformer head. For current transformers with a hairpin primary winding the location of the arc is chosen at
the bottom of the transformer, more precisely, in the transformer tank. For voltage transformer, the area of the maximum electric
field is proposed. Other technical requirements were not changed in relation to the initial ones.
The next proposal variant from 2005 required a test current of 60–80% (Ith), with a dynamic current 1.7 times higher. Test
requirements and criteria are given in Table III.
The location of arc inception in SF6 transformers was the area with the highest electric stress. For current transformers with
paper–oil insulation, the location of arc inception was the same as in the proposal in 2004, but this time the bottom of the tank was
proposed as the arc location for voltage transformers.
The proposal was modified again at the end of 2005 at the general meeting of IEC TC 38 in Opatija. Since the explosion of a
transformer in service that has been tested and rated ‘explosion proof’ can also have legal consequences, categories ‘internal arc
fault protection—instrument transformers class I or II’ are proposed in accordance with the American practice. Class I replaces the
‘explosion resistant’ transformer, and class II the ‘explosion proof’ one. Additionally, for instrument transformers with paper–oil
insulation it is proposed that the location of arc initiation should be agreed to between the buyer and the manufacturer.
From the above information it can be concluded that these frequent, and sometimes contradictory changes of the fundamental
proposal are a result of numerous controversies. What is controversial in this proposal?
Firstly it is the pull back from the basic intention to attain high in service reliability for instrument transformers. Instead of that,
especially for paper–oil insulated instrument transformers, attempts are made to find a compromise and define the minimum of test
modes and requirements in order to enable some transformers to pass the internal arc tests.

Table III. Internal arc duration and criteria for instrument transformer testing.
Arc Degree of Arc Explosion Explosion
Current protection duration proof resistant
<40 kA 1 0.2 No external effects except operation of arc relief device Fracture of enclosure and insulators,
2 0.5 No damaged parts (fire acceptable) and fire acceptable, but broken
components must be within the
40 kA 1 0.1 No external effects except operation of pressure prescribed radius
relief device
2 0.3 No damaged parts (fire acceptable)

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
REDUCTION OF IN-SERVICE INSTRUMENT TRANSFORMER EXPLOSIONS 931

Figure 1. Location of the most stressed area of an inverse-type instrument transformer (a) and results of electrical field computation (b).

As well the level and the dynamic component of arc current are controversial. The most probable value of the fault current in
power systems during instrument transformers breakdown should be checked by experiments.
The proposed location of arc inception in instrument transformers with paper–oil insulation is definitely wrong. According to the
proposal, only faults within metal enclosures are in part covered by the test. As it is known, breakdowns or arcing in insulating
systems can occur between any two electrodes at different potentials, and here the rule applies that the most probable breakdown
location will be at the highest dielectric stress area.
In high-voltage instrument transformers that is as a rule always on the edges of potential electrodes inserted in the insulation. As
shown in Figure 1 for the inverse-type current transformer (red circle) that place is always inside the insulator.
The statement that the most probable location of the fault occurrence is at the transformer top (designated with K in Figure 1.) and
emphasised in the draft of the standard, is the result of experience of some transformer users. In the authors’ opinion, this position is first of all
taken due to design and technological errors by particular manufacturers of instrument transformer, and can be relatively easily eliminated.
To prove this assumption, a severe duty cycle of dielectric tests was performed on a 123 kV current transformer from regular
production, as shown in the diagram in Figure 2.

Figure 2. Test diagram of 123 kV current transformer under the condition of extreme dielectric stress.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
932 M. POLJAK AND B. BOJANIĆ

Figure 3. Damaged transformer after breakdown (a) and location of fault arcing formation (b).

The voltage of 184 kV (80% of rated power-frequency withstand voltage) was first applied to the transformer for 24 hours. In the
next 24 hours 100% of the test voltage was applied, and after that 250 and 275 kV for 5 hours each. Finally, there was a breakdown
after 20 minutes of application of 300 kV. Test effects are shown in Figure 3.
Figure 3 shows beyond any doubt that if instrument transformer insulation is correctly dimensioned, and if there are no hidden
technological or other errors, faults due to voltage will occur at the location of the highest electric field, where arcing will be
initiated. To expect that, under the conditions similar to the one described in the above test, a transformer will satisfy the ‘explosion
proof’ criteria is technically mission impossible.
Therefore, instead of ‘explosion proof’ concept the authors propose the ‘explosion safe’ concept.

4. ‘EXPLOSION SAFE’ INSTRUMENT TRANSFORMER

Instead of the concept and strategy based on IEC proposal, the authors propose a new concept that improves operating reliability.
The new concept includes tests and measures that, if performed, reduce in theory, the possibility of an explosion.
An ‘explosion safe’ transformer is one:
(1) whose partial discharges do not exceed 10 pC at rated power-frequency withstand voltage (routine test);
(2) whose partial discharges, after switching, lightning and multiple chopped impulse voltage tests, do not exceed 10 pC at rated
power-frequency withstand voltage, and the increase in the concentration of gasses dissolved in oil does not exceed defined
values (special test) and
(3) is equipped with an overpressure gauge or a similar device.

4.1. Ageing tests of electrical insulation


The theoretical foundation for the first criterion can be found in investigations of behaviour of instrument transformer insulation
under the influence of an electric field, References [4] and [5], where it is shown that the probability of occurrence of partial
discharges can be represented by a Weibull distribution.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
REDUCTION OF IN-SERVICE INSTRUMENT TRANSFORMER EXPLOSIONS 933

Figure 4. Main dimensions of the models.

The tests were performed on several different models to investigate the effects of electrical field on partial discharge inception
voltage.
The insulation system of the models consisted of two cylindrical capacitors in series wound on an aluminium tube (Figure 4). The
models were dried, subjected to a vacuum and impregnated together with 123 kV instrument transformers. Other parameters that
affect life of instrument transformers, e.g. temperature and moisture, were not taken into consideration.
Each sample was subjected to an increasing voltage up to a predetermined level until partial discharge inception occurred or for
the maximum of 1 hour if there was no partial discharge. Experience shows that such tests are non-destructive, and that after a
certain time from de-energisation the insulation resumes its original properties [4]. The partial discharge inception voltage is the one
at which partial discharges exceeding 10 pC with a duration of at least 15 seconds appear. Test voltages were chosen in accordance
with preliminary measurements or previous investigations. The choice is made in the way as to have levels with small, medium or
high probability of partial discharge inception.
Depending on whether or not partial discharges occurred at a certain voltage, the next test value will be one level higher or lower.
For each voltage level the test results were presented separately as a function of elapsed time until partial discharge inception. To
simplify the processing of results, these values were grouped into several chosen time intervals. The number of partial discharge
inceptions Ni, which occurred in each of such chosen intervals expressed as a percentage of the total number of tests N at a certain
voltage level, gives the empirical probability P of partial discharge inception in percent (Table IV).
In 1939 Swedish scientist Waloddi Weibull [6] introduced a distribution function, which very often gives an excellent statistical
description of failure theory, failures, residual life assessments and the like.
Statistically, the cumulative distribution function, i.e. the probability P, can be calculated with Equation (1)

P ¼ 1  e½f ðuÞt (1)

where f (u) is the function dependent on the applied voltage. If in the Expression (1) the exponent is not linear with time, we are
talking about the ageing effect. The expression for probability P which includes both physical and statistical aspects of insulation
systems is now:
a tb
P ¼ 1  eAU (2)

where A, a and b are constants.

Table IV. Results of tests on models with radial width 2  0.75 mm.
U/kV N t < 0.5 min t < 3 min t < 10 min t < 35 min t < 60 min
Ni P/% Ni P/% Ni P/% Ni P/% Ni P/%
20 17 0 0.00 1 5.88 1 5.88 1 5.88 1 5.88
22 52 4 7.69 6 11.54 7 13.46 13 25.00 15 28.85
24 60 10 16.67 15 25.00 24 40.00 32 53.33 38 63.33
26 25 10 40.00 13 52.00 15 60.00 22 88.00 23 92.00
28 4 1 25.00 2 50.00 3 75.00 4 100.0 4 100.0

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
934 M. POLJAK AND B. BOJANIĆ

For an analysis of measurements of partial discharge inceptions, results from Table IV can be expressed with Equation (3)
P ¼ f2 ðUÞt¼c (3)

Using the above expression, it is possible to obtain the relationship between voltage U and time t for the given probability P, i.e. to
obtain volt–second curves, which describe the time dependence of various physical phenomena connected with applied voltage.
Typical examples are partial discharge inceptions and insulation breakdowns.
One of the main problems of Weibull distribution is the determination of its parameters. Combining Expressions (2) and (3), and
assuming that Atb ¼ k then
  
1
ln ln ¼ a ln U þ ln k (4)
1P

In the Weibull probability diagrams, Equation (4) describes the line


Y ¼ aX þ ln k (5)

For the given voltage U and probability P from Table IV, values of Y and X for parameter t are computed and imported into an
Excel spread sheet. The sought-for straight line, i.e. parameters a and ln k, are obtained by linear regression using the method of least
squares (Figure 5). Besides Weibull curves, equiprobabilistic the volt–time curves (curve with constant value of P), are also
interesting and important for practical applications and analyses.
If in Equation (2) P ¼ constant, then the expression for volt–time curve:
ln U ¼ ln C  g ln t (6)
where ln C ¼ 1=aðln ln 1=ð1  PÞ  ln A i g ¼ b=a
Expression (6) can also have the form:
U ¼ C=tg (7)

It can be shown that the exponent g is practically the same for all the probabilities, so that volt–time curves can be defined by
relative values. Reference voltage corresponds to the reference time t ¼ tR , and its percentage (8) corresponds to any other time t.
U% ¼ 100ðt=tR Þg (8)

The exponent g can be determined also by Expressions (9):


lnðU1 =U2 Þ
g¼ (9)
lnðt2 =t1 Þ
where (U1, t1) and (U2, t2) are two points of volt–time curves with the same probability.

Figure 5. Weibull curves for models with radial width 2  0.75 mm.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
REDUCTION OF IN-SERVICE INSTRUMENT TRANSFORMER EXPLOSIONS 935

Figure 6. Absolute values of volt–time curves for the tested models.

Figure 6 gives absolute values of partial discharge inception voltages versus time (volt–time curves) for all three models, while
Figure 7 shows percentage volt–time characteristics of models of oil–paper insulation, which are typical for instrument transformers
and which are extrapolated for increased time. Partial discharge inception probability curve decreases with time, and can be
expressed by exponential law. Exponents g are from 0.035 to 0.049.
In the Reference [4] for insulating configurations typical for power transformers that coefficient ranges from 0.019 to 0.059.
The obtained results show that the exponent g increases with thickness of insulation between potential screens, which suggests
that in practice it is more useful to use an insulation system with more screens.
The practical application of these investigations will be shown on the model with the thickest insulation (red curve). On the basis
of Expression (8) or Figure 7, it can be shown that if an insulation has withstood a certain voltage for 1 minute without partial
discharge, then with the same probability it will endure without partial discharge for 2.97 years (1.56  106 min) if half of that
voltage is applied. In high-voltage networks the ratio between rated voltage and rated power-frequency withstand voltage is from
0.28 to 0.37. In same way, it can be shown that if a transformer has had no partial discharges during the power-frequency withstand
voltage test, its insulation will remain satisfactory for thousands of years.
The influences of temperature and moisture were not taken into account in these investigations. Further, it must be stressed that
the maximum duration of tests was limited to 60 minutes, so that extrapolation to some longer period can be questionable. In papers
dealing with experiments on power transformer insulation [4] it has been shown that extrapolation is not questionable to 104 min,
equalling 7 days. Making conclusions for a longer time period without prior experimental checks is acceptable only with difficulty;
simply because the mechanism of partial discharge inception is not adequately known, and it can vary with time and type of

Figure 7. Volt–time curves for models of instrument transformer insulation.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
936 M. POLJAK AND B. BOJANIĆ

modelled insulation [7]. Testing and measurements to the point of a dielectric breakdown on a complete 123 kV current transformer,
described in former chapter qualitatively demonstrate that extrapolation could be acceptable.

4.2. Other criteria for ‘explosion safe’ transformer


The second criterion in the definition of the ‘explosion safe’ transformer is based on the fact that partial discharge in service can be
caused by lightning, switching and fast transient caused overvoltages. Absence of partial discharge after their laboratory simulation
is a reliable criterion that defines transformer safety in service. Additionally, after those tests it is possible to measure the content of
dissolved gases in the transformer oil, which raises the sensitivity of the test procedure to a very high level.
The third criterion in the definition of the ‘explosion safe’ transformer has already been dealt with in the previous sections. Let us
state here that a device for monitoring the transformer oil pressure increase will be an efficient means of incipient failure detection in
the event of mechanical damage of the insulation, moisture ingress into the insulation and similar occurrences which all lead to
slowly developing transformer failures. CIGRE survey indicates that approximately 90% of failures are of that nature.

5. CONCLUSION

On the basis of research performed on models, a systematic analysis of the CIGRE survey of instrument transformer failures, and
tests performed on an inverse-type current transformer for Um ¼ 123 kV, the IEC approach whose aim is to prevent in-service
explosions of instrument transformers is critically disputed. Instead of direct investigations of internal arc effects, which are in many
respects not adequate to explain actual failures, a new way of improving the in-service reliability of instrument transformers is
proposed. The described investigations and experience from service show that the reliability of the main insulation can be
objectively verified in the factory by routine, type and special tests. The remaining problems are related to winding faults and the
resulting damage due to short-circuit, transport, moisture, leakage and the like. These faults slowly generate gases, resulting in an
increase of the oil pressure. If transformers would be equipped with a sensor or some similar device for the indication of the oil
pressure increase, it would be possible to prevent almost 90% of serious instrument transformer failures and incidents.
Therefore, the new concept called ‘explosion safe’ consists of:
- measurements of partial discharges at rated power-frequency withstand voltage (routine test);
- measurements of partial discharges after tests with multiple chopped, lightning, and switching impulse voltages;
- application of sensor for oil pressure measurement, or some similar device.
This approach is fully applicable to instrument transformers with paper–oil insulation, and its acceptance would eliminate the
necessity of expensive and destructive tests. The main assumptions, on which this concept is based, have been verified both in theory
and practice taking into account the principles of the Weibull statistical distribution.
When talking about the ‘explosion safe’ transformer, the kind of insulators used is the decisive factor for the consequence of an
internal arc. Transformers with silicon insulators are considered very safe and very reliable with regard to explosions in service.
As to instrument transformers with SF6 insulation, the proposed concept is not appropriate and inapplicable to them. Due to gas
compressibility, it is practically possible to avoid, through correct electrical and mechanical dimensioning, explosions and damages
even in the cases when an arc is formed inside the porcelain insulators. It is inherent in SF6 insulated instrument transformers that, in
the case of arc in the transformer, the protective pressure relief device will activate, and that gas will be vented to the atmosphere.
This is a regular experience in laboratory tests and tests performed in accordance with the new IEC proposal [8]. Because of the
considerable greenhouse effect, the venting of SF6 gas is considered a serious ecological incident in many countries today.
Therefore, the IEC proposal also imposes on laboratories the requirement to find ways to avoid that.

6. LIST OF SYMBOLS

g exponent
a, b, k constants of Weibull distribution
C coefficient
P probability
t time

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep
REDUCTION OF IN-SERVICE INSTRUMENT TRANSFORMER EXPLOSIONS 937

tR reference time
U voltage in time–voltage curve
(Ui, ti) point of volt–time curves with the same probability
X abscissa of Weibull straight line
Y ordinate of Weibull straight line

REFERENCES

1. CIGRE Brochure, The paper-oil insulated measurement transformer, CIGRE, Working Group 07 of SC 23 (Substation), 1990.
2. Poljak M, Bojanić B. Instrument transformers—Diagnostics as precondition of reliable service, Almanac of 5. International symposium ‘Diagnostics electrical
machines, transformers and apparatus’, Rovinj 2000, Paper T1, Croatian.
3. Lemaitre F. Comments on documents TC38/MT30/0126 and 0127, Internal arc fault requirements and tests for current and inductive voltage transformers.
4. Yakov S. Volt-time relationships for PD inception in oil-paper insulation, ELEKTRA, 1979; 17–28.
5. Poljak M. Insulation system of combined instrument transformers, Ph.D Thesis, FER-Zagreb, 2006, Croatian.
6. Hallinan AJ. A review of the Weibull distribution. Journal of Quality Technology 1993; 25(2):85–93.
7. Brown GW. The Weibull distribution: some dangers with its use in insulation studies. IEEE Transection on PAS 1982; PAS-100(9):3513–3522.
8. IEC 61869-1, Instrument Transformers—Part 1: General Requirements,1st edn, 2007, IEC standard.

Copyright # 2009 John Wiley & Sons, Ltd. Euro. Trans. Electr. Power 2010; 20:927–937
DOI: 10.1002/etep

You might also like