Paper Published on the16th International Symposium on High Voltage Engineering, Cape Town, South Africa, 2009

UNCERTAINTIES IN THE APPLICATION OF ATMOSPHERIC AND

ALTITUDE CORRECTIONS AS RECOMMENDED IN IEC STANDARDS
Dong Wu1*, Ming Li2 and Mats Kvarngren3
1
ABB HVDC, SE-771 80 Ludvika, Sweden, 2ABB Corporate Research, SE-721 78
Västerås, Sweden, 3STRI, SE-771 80 Ludvika, Sweden
*Email: [email protected]

Abstract: The dielectric strength of air is influenced by air density (temperature and pressure) and
humidity. Such effects need to be taken into account when external insulation is designed and
tested. Since the conditions of application and the conditions of the laboratory tests may be
different, it is often necessary to make corrections between different atmospheric conditions. For
engineers at manufacturers, utilities or high-voltage laboratories, they follow the relevant IEC
standards. However, atmospheric conditions influence the dielectric strength of air in a
complicated way. Simplified and generalized solutions may cause vacillations especially when
different recommendations are given in different standards without sufficient clarifications. It is
the intension of this paper to give an outline of such issues that may lead to uncertainties in the
application of various IEC standards regarding the atmospheric correction. Some proposals are
also given for discussion.

1. INTRODUCTION Where: p0 and t0 (in degree C) are the pressure and

temperature at the standard reference conditions
The dielectric strength of air is influenced by air respectively; p1 and t1 are that at other air conditions.
density (temperature and pressure) and humidity. Such
effects need to be taken into account when external Thereafter the atmospheric conditions are converted
insulation is designed and tested. Since the conditions into mainly two parameters, the relative air density and
at the application and the conditions at the laboratory the absolute humidity. These two parameters have also
tests may be different, it is often necessary to make certain co-effects. It should be noted that for outdoor
corrections between different atmospheric conditions. conditions, it might be assumed that the effects of
Creditable studies and reviews have been published on ambient temperature and humidity tend to cancel each
the atmospheric corrections, e.g. [1, 2]. These studies other [4].
were the base for the recommendations in IEC
standards, e.g. [3-10]. For engineers at manufacturers, 2.2. Influence of air density
utilities or high-voltage laboratories, they follow the
The breakdown of a non-uniform long air gap takes
relevant IEC standards. However, atmospheric
often the processes as corona inception, streamer
conditions influence the dielectric strength of air in a
propagation, leader formation and propagation, and
complicated way. Simplified and generalized solutions
final jump. The streamer and leader processes are the
may cause vacillations especially when different
recommendations are given in different standards decisive processes. It has been concluded in literature
without sufficient clarifications. [1-2] that the influence of air density is most
significant on the streamer formation and propagation.
The air density has little influence on the leader
Based on practical engineering experience, it is the
process. Therefore, as an approximation, one may
intension of this paper to give an outline of such issues
consider if the streamer dominates the breakdown
that may lead to uncertainties in the application of
processes in a gap, the dielectric strength of this air gap
various IEC standards regarding the atmospheric
is proportional to relative air density. This is in
correction. Some proposals are also given for
principle the case for shorter gaps, shorter than 2
discussion.
meters. For longer gaps, the breakdown will be resulted
by both the streamer and the leader process. Therefore,
2. CORRECTION FOR AIR INSULATION
the dielectric strength of a longer air gap is, in many
cases, less than proportional to air density.
2.1. Related parameters of air
The dielectric strength of air is influenced by the air According to above approximations, the change of the
density (temperature and pressure) and humidity. The dielectric strength of air gaps with air density may be
influence of temperature and pressure can be taken into evaluated by the value of δm. For breakdown caused
account simultaneously, at least as a first mainly by streamers, m is equal to one (m=1).
approximation, by the relative air density, δ, [2]: Otherwise, for most of the non-uniform long gaps, m is
smaller than one (m<1). Therefore, the main task for
p1 273 + t0 air density correction is to evaluate m. However, the
δ= × (1)
p0 273 + t1 value of m is not just a simple function of gap length, it
is also the function of gap structure, voltage type, as
well as the relative air density, δ, in a complicated way. applied with an acceptable accuracy for cases where
We have m=f(U/L, kgap, δ). Here kgap is the gap factor. 0.9≤δ×k≤1.1 [2]. Such limitation may only be fulfilled
when correction is made between test results obtained
2.3. G factor method near sea level. For high altitude correction other curves
give better accuracy. This is an issue being overlooked
Based on creditable studies and test results obtained at by this standard.
altitude over 3000 meters, the correction methods using
“G factor” were summarized and presented [1-2]. A In this standard, it is recommended to use the 50%
factor, G, is introduced to simplify the complicated breakdown voltage and the minimum discharge path to
relation of m=f(U/L, kgap, δ) by a semi-empirical evaluate G. When U50 is not available, Ub can be
approach m=f(G). The value of G is the ratio of the assumed to be 1.1 times of the voltage level of the
mean electric field, E, at the breakdown voltage of a withstand test. However, such assumption can only be
given gap, and the average electric field of the positive justified when the voltage level of the withstand test is
streamer, Es, at the same atmosphere. related to the value of U50 and the minimum discharge
E path in such a way that it reflects the discharge
G= (2) characteristics of the test object. Otherwise, errors will
Es be introduced to the value of E=Ub/L and thereafter to
G and m. For the purpose of insulation coordination
With E=Ub/L, Es=Es0×δ×k, and Es0=500 kV/m.
and the determination of type test voltage, the value of
In the above relations, Ub is the breakdown voltage of U50 and the length of related minimum discharge path
the air gap and L is length of the discharge. k is the are often not available. The test voltage specified has,
correction for humidity; and Es0 is the average electric in many cases, a weak relation with an arcing distance
field of the positive streamer at standard reference of the equipment. In such a case, E=Ub/L can not be
atmosphere. It is intended to determine, by the value of correctly estimated as recommended by this standard.
G, how much a breakdown will be contributed by
streamer at the given gap. G value is actually the 3.2. IEC 60071-2 (1996-12)
simplified presentation of the breakdown characterises
In this standard, [4], the altitude correction is given in a
of the given gap. The functions of m=f(G) is given
formula as:
Figure 1, with the value δ×k as index.
H
m
ka = e 8150
(4)

m Where: H is the altitude in meters. Actually, we can

relate ka in formula (4) with kt in formula (3) by:
m
ka=1/kt=1/δ .

Figure 1: The relations of m=f(G) given in [2].

3. RECOMMENDATIONS IN IEC
STANDARDS

3.1. IEC 60060-1 (1998-11)

In this standard [3], the correction factor, kt, is defined
as:
m
kt = δ ×kw= U/U0 (3) Figure 2: The relations of m=f(Ucw) given in [4]
Where U is the breakdown voltage at given site
conditions and U0 is the breakdown voltage at standard Instead of using the relation m=f(G), as that in IEC
reference conditions. G factor method is adopted in this 60060-1 [3], m is given here as the function of the co-
standard. Both m and w are the function of G. ordination withstand voltage, i.e., m=f(Ucw). This is a
simplified and conservative approach, avoiding the
In this standard, the curve with solid line in Figure 1, difficulties in obtain the relation of E=Ub/L. This is
for δ×k=1, is used, for m=f(G). This curve may only be
very convenient for the purpose of insulation co- 3.7. IEC 60168 (2001-04)
ordination and the determination of type test voltage
In this standard [10], no recommendation is given for
altitude correction. If the atmospheric conditions at the
Different values for m are recommended for different
time of test differ from standard reference atmosphere,
types of voltage and insulation. For lightning impulse
then corrections shall be made according to IEC
voltage (LI) and short duration AC voltage, m=1 is
60060-1[3].
recommended. For switching impulse voltage (SI), the
value of m is to be obtained through a group of curves
4. DISCUSSIONS AND PROPOSALS
in Figure 2.
4.1. Application of IEC 60060-1
3.3. IEC 62271-1 (2007-10)
The atmospheric correction recommended by this
In this standard [5], the normal service conditions
standard [3] is most accurate when breakdown tests
including altitudes not exceeding 1000 meters are
have been performed on air gap or dry insulators. In
specified. The altitude correction recommended is said
such cases, the dielectric characteristics of the test
to be applicable to 4000 meters with formula below:
objects can be obtained. The value of E=Ub/L can be
m
H −1000 evaluated accurately.
ka = e 8150
(5)
This correction becomes difficult to apply when
This is the same formula as formula (4) but correction is to be made on withstand test voltage of
avoided the correction for the altitude not equipments. This is because in some cases, the relation
exceeding 1000 m. For SI, further simplifications between the withstand voltage and U50 is not 1.1 times.
and approximations are made. Instead of the In some cases the discharge path may not be
curves in the form of m=f(Uw), as given in Figure determined by the voltage level of this test but by other
2, constant values of m for different type of constrains, such as creepage or installation
requirement. In some cases, the shortest discharge path
voltages are recommended. The correction is not
is not the critical insulation under this test voltage. The
anymore the function of voltage level: value of E=Ub/L in such cases may not be correctly
• For AC, LI, and phase to phase SI: m=1 evaluated. Even if one may introduce certain type of
iterative calculation procedure between the test voltage
• For Longitudinal SI: m=0.9 and the correction factor, the uncertainty in evaluate
• For phase to earth SI: m=0.75 E=Ub/L can not be solved. However, as long as the
corrections are made for relative small difference in air
density from standard reference conditions, i.e.,
0.9≤δ×k≤1.1, the magnitude of the error introduced
3.4. IEC 60076-1 (2000-04) and IEC 60076-3
may still be tolerated.
(2000-03)
In these two standards for power transformers [6-7], The extended application of the relation m=f(G) in this
the normal service conditions including altitudes not standard to high altitude correction will lead to a
exceeding 1000 meters are specified [6]. The altitude increased error. The correction as recommended today
correction is made directly on the gap distance, i.e. “if is not suitable for altitude correction. One solution
the transformer is specified for operation at an altitude would be to adopt the complete relation of m=f(G) as
higher than 1000 m, the clearance requirements shall that given in Figure 1 into this standard. The other
be increased by 1% for every 100 m by which the solution is to use the recommendations in IEC 60071-2
altitude exceeds 1000 m” [7]. for altitude correction, limiting the application of this
standard in the range of 0.9≤δ×k≤1.1.
3.5. IEC 60137 (2008-07)
In this standard [8], it is said that “bushings 4.2. Application of IEC 60071-2
corresponding to this standard are declared suitable for
operation at any altitude not exceeding 1000 m”. The The atmospheric correction recommended by this
altitude correction recommended is applicable to 4000 standard is most suitable for the purpose of insulation
co-ordination and the determination of type test
meters with formula (5). Constant values for m are
voltage. Since the correction factor is related directly to
also recommended as: the co-ordination withstand voltage.
• m =1 for power frequency and LI;
One importance issue in this correction is whether or
• m =0.75 for SI. not to make correction for the altitude not exceeding
3.6. IEC 60044-8 (2002-07) 1000 meters. On this point this standards differs from
many other IEC standards [5-9]. In this discussion,
This standard [9] has exactly the same approach for the there are actually four different atmospheric conditions
altitude correction as that in IEC 62271-1 [5]. in the context.
 1. Standard reference conditions with conditions, e.g., at an altitude higher than 1000 meters,
temperature of 20oC, air pressure of 101.3 atmospheric corrections between these two conditions
kPa, and absolute humidity of 11g/m3 are necessary for both the design and test of the
external insulation. For equipment which will be tested
2. Normal service conditions (conditions that
at a laboratory where conditions at the day of testing
specified for various HV equipment in
differ from standard reference conditions, atmospheric
relevant standards) with maximum ambient
correction is necessary for the test voltage between
temperature of, e.g., 40oC, altitude not
these two conditions.
exceeding 1000 meters, and …
3. Specific site conditions (application Taking this discussion into account, it is justified to use
conditions) with altitude of, e.g., 1600 meters, formula like that in IEC 62271-1, i.e., (5) instead of
and… formula in IEC 60071-2, (4).
4. Laboratory test conditions (at the day of
4.3. Values of m for different types of voltage
testing) with ambient temperature of, e.g.,
25oC, air pressure of, e.g., 100.0 kPa and In several IEC standards, the values of m for different
relative humidity of, e.g., 40%. voltage types are recommended.
These different conditions represent different severities
The recommendation of m=1 for LI is justified. The
for the external insulation design and test. In Figure 3,
mean field of a rod-plane gap under positive LI is
the relations between the different atmospheric
conditions are sketched out. higher than 500 kV/m, i.e., E=Ub/L≥500 kV/m.

The recommendation of m=1 for AC is a conservative

Severity Specific site conditions approach. The non-linear characteristic of the AC
breakdown voltage against gap length indicate the
kt same breakdown process as that appears under SI [2].
Theoretically kt For long gaps, there will be m<1. On the other hand, a
rod-plane gap of 2 meters will have a U50 for short
Normal service conditions duration AC of 620 kV. This voltage corresponds to a
phase to phase voltage of 1074 kV. Therefore, if AC is
Standard reference conditions kt
Lab. test conditions the dimensioning voltage, with the system voltage
available today, relative short gaps will be required.
Taking this into consideration, the conservative
approach of using m=1 may be justified. Otherwise for
Figure 3: Atmospheric correction between different longer gaps, the value of m should be similar to what is
ambient conditions used for SI. Therefore, the recommendations would be:

Theoretically, for the design and test of external • For gaps of lengths shorter than 2 meters, m=1
insulation, atmospheric corrections should be applied
between the specific site conditions, laboratory test • For gaps of lengths longer than 2 meters, use
conditions and the standard reference conditions. the same m value as that for SI obtained in
However, for most equipment, there are normal service Figure 2
conditions specified in the relevant IEC standards. For
Note that peak AC voltage should be used as SI in
economical reasons and industrial practice, equipments
Figure 2.
have to be designed to withstand the required withstand
voltage within the range of normal service conditions.
No correction for DC was included in most of these
The differences between the standard reference
standards. In Cigré report a linear relation between the
conditions and the normal service conditions have been
short duration DC breakdown voltage and the length of
included in the design. This means that for equipment
rod-plane gap has been reported up to 1000 kV with
used at a location with altitude not exceeding 1000
mean breakdown field in the level of 500 kV/m [2].
meters; no altitude correction will be necessary. This
Since the gap length is short, m=1 would be used.
industrial practice has been supported by vast
However, other literature has reported slightly non-
operational experience and adopted by many IEC
linearity from 2 to 4 meters of gap lengths [11]. The
standards. This is especially true for the cases when
mean breakdown field is in the level of 400 kV/m. This
deterministic method is used for insulation
will result in: G=0.8 and m=0.6. More laboratory
coordination, e.g. for HVDC systems (IEC60071-5,
studies are needed for longer gaps. However, with the
under revision). The effect of the air density in this
limited results today, the recommendations would be:
range, where H 1000 meters, has already been
included in the margins commonly adopted for
insulation design. • For gaps of lengths shorter 2 meters, m=1
• For gaps of lengths longer than 2 meters,
For equipment that will be used at the specific site m=0.6.
conditions more severe than the normal service
The recommendation of use m=0.75 for all SI level is a 2. The correction method recommendation in
conservative approach. This value may be justified for IEC 60071-2 is convenient to use for purpose
EHV systems. For UHV system, relations in Figure 2 of insulation co-ordination and the
should be followed. determination of withstand test voltages. It
can be applied for altitude correction for
4.4. Altitude correction for creepage altitudes up to 4000 m, as being recommended
by several other IEC standards, before any
Along hydrophilic surfaces, when pollution has been research results prove otherwise. It is a
wetted, dry-band related discharge activities may take
conservative approximation of G factor
place. This is a short gap with streamer breakdown.
method.
The dry-band can become wetted again after the
current flow though the arc in air. Such activities can
3. For equipment with a specified normal service
occur several times and a full flashover of the insulator
conditions including altitudes not exceeding
may take place. In this process, the change of air
1000 m, altitude correction should only be
density will change the dielectric strength of the air. If
applied to correct the altitude exceeding 1000
it was pure air breakdown in the form of streamer, then
m as that given in formula (5):
the correction would be kt=δm with m=1. However, the
conditions of both surfaces in series and parallel to the H −1000
dry-bend activity contribute to the effects. For real m

insulator, the change of pollution level and the change

ka = e 8150

of creepage distance in parallel with the air gap will

change the interaction between the air gap and surface 4. the recommended value for m is proposed to
activities. The insulator shed profile play also an be related to gap length as:
important role. This is again a complicated
phenomenon [12]. • For LI: m=1.
• For SI: using curves in Figure 2.
Although wide differences exist in literature, it is
recommended that a constant value of m should be • For AC: when gap length is shorter than 2
used with m=0.5 for AC and 0.35 for DC [12]. The meters, m=1. When gap length is longer
correction is to be applied for voltage, the same as in than 2 meters, use the same value of m for
equation (3): SI with the same voltage level to AC peak
voltage.
kt = U/U0=δm
• For DC: when gap length is shorter than 2
It is the same as for air insulation correction; no
meters, m=1. When gap length is longer
correction for the creepage distance is needed for
than 2 meters, m=0.6.
altitude up to 1000 meters. In pollution test, for the
insulators of the same shed profile and at a given 5. Altitude correction for creepage should be
pollution level, the relationship between the U50 and the introduced into IEC standard as that
creepage distance is in most cases linear [12]. recommended by Cigré Review [12].
Therefore, one can replace the voltage in equation (3)
with creepage distance, L: 6. REFERENCES
-m
kc = L/ L0 =δa (6)
[1] K. Feser, A. Pigini, ”Influence of atmospheric
Here L is the creepage distance for a high altitude conditions on the dielectric strength of external
and L0 is the creepage distance for altitudes up to insulation” Paper prepared at the request of the
1000 meters. Chairman of SC 33, Electra No. 112.
[2] WG 33-07, “Guidelines for the evaluation of the
5. CONCLUSIONS dielectric strength of external insulation” Cigré
Brochre 72.
For the recommendations in IEC standards on the [3] IEC 60060-1, High-voltage test techniques – Part
atmospheric and altitude correction the following may 1: General definitions and test requirements.
be considered. Edition 2.0, 1989-11
[4] IEC 60071-2, Insulation co-ordination – Part 2:
1. The correction method recommendation in Application guide. Third edition, 1996-12
IEC 60060-1 is most accurate for breakdown [5] IEC 62271-1, High-voltage switchgear and
test performed at the condition where controlgear – Part 1: Common specifications.
0.9≤δ×k≤1.1. It is inaccurate when used for Edition 1.0, 2007-10
high altitude correction. It is inconvenient and [6] IEC 60076-1, Power transformers – Part 1:
inaccurate when used for insulation co- General. Edition 2.1, 2000-04
ordination and the determination of withstand [7] IEC 60076-3, Power transformers – Part 3:
test voltages. Insulation levels, dielectric tests and external
clearances in air. Second edition, 2000-03
[8] IEC 60137, Insulated bushings for alternating
voltage above 1000 kV. Edition 6.0, 2008-07
[9] IEC 60044-8, Instrument transformers – Part 8:
Electronic current transformers. First edition,
2002-07
[10] IEC 60168, Test on indoor and outdoor post
insulators of ceramic material or glass for systems
with nominal voltage greater than 1000 V. Edition
4.2, 2001-04
[11] S.J. Huang, Z.H. He, Z.B. Wen, M.G. Wang,
“Flashover tests on large air gaps with DC
voltage”, ICPST’94 Beijing, China
[12] Taskforce 33-04-01: “Polluted insulators: a review
of current knowledge” Cigré Brochre 158