Electric Power Systems Research 165 (2018) 94–101

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Rules to estimate the expected values of zero-sequence impedances in 3- T

phase core-type transformers
⁎
Elmer Sorrentinoa, , Juan Carlos Burgosb
a
Universidad Simón Bolívar, Caracas, Venezuela
b
Universidad Carlos III de Madrid, Leganés, Madrid, Spain

A R T I C LE I N FO A B S T R A C T

Keyword: Simple rules to estimate the expected values of zero-sequence impedance measurements for 3-phase core-type
Transformer zero-sequence impedances transformers are shown. The proposed rules are based on the analysis of magnetic circuits. Due to this fact, a
conceptual description about zero-sequence impedances of these transformers is included. The rules were ver-
ified with measurements on fifteen transformers, and these measurements were used for refining the rules. In
case of magnetizing zero-sequence impedances, the presence or not of magnetic shunts on tank walls defines the
details of these rules. In case of short-circuit zero-sequence impedances, only the positive-sequence transformer
impedances are necessary to apply the proposed rules. These rules are important in order to reduce the prob-
ability of errors during the tests. Some rules related to the resistive part of the impedances are also presented.

1. Introduction Most of available values for this article were taken from measure-
ments on YNynd transformers. The analysis of YNynd transformers re-
Three-phase core-type transformers are also known in the literature quires the study of YNyn and Dyn connections. For 3-phase core-type
as 3-phase 3-limb transformers, or 3-phase 3-leg transformers. Zero- YNynd transformers: (a) Z0M are measured with the other wye side in
sequence impedances (Z0) of 3-phase core-type transformers have been open-circuit and the tertiary in open-delta condition; (b) all the other
studied for years [1–20]. There are standardized procedures for Z0 possible measurements are Z0SC, and they can be obtained with one
measurements [11–13], which are based on feeding a wye side of the transformer side without current or with all the windings with currents.
transformer with zero-sequence currents. Induced zero-sequence cur- These differences between the Z0 values are not evident in some
rents could be or not circulating in other transformer sides, depending literature about this topic. There are references with a clear difference
on the specific test. Only the standardized procedures for Z0 measure- between the diverse Z0 values [1–13], but some documents only present
ments are considered in this paper. a wide range of “reactor-type” values [14–16]. The relative position
Z0 values of transformers depend on their design. Measured values (inner/outer) of the winding connected to the source during the test
of Z0 for transformers may be classified as [1]: “no-load type” (very determines the values of the measurements in core-type transformers,
high value), “reactor type” (Z0 is in the order of 1 pu), and “short-circuit and only few references [6–11] consider this key point, but useful ways
type” (Z0 values are relatively low, in the same order of positive-se- to estimate the measurements are only available in a subset of them
quence short-circuit impedances). “No-load type” and “reactor type” [10,11]. Fortunately, an IEC standard [11] has some rules for esti-
values are measured with current only in the winding connected to the mating the values of Z0 measurements, but without any explanation
source (under these conditions, 3-phase core-type units have “reactor about the origin of such rules. Some concepts related to the nature of
type” values, whereas 3-phase units with closed ferromagnetic path for these Z0 values were previously analyzed (in papers about their de-
zero-sequence fluxes have “no-load type” values). For 3-phase core-type termination, with the help of computing magnetic fields in transformer
transformers, “reactor type” values can be also called zero sequence geometry [21–23]), but an integral description about them is still ne-
magnetizing impedances (Z0M) because there is not circulating current cessary, and this paper is a contribution for it.
in the other windings (but there is induced zero-sequence current in the Unavailability of expected values for Z0 has an influence in some
tank). Zero-sequence short-circuit impedances (Z0SC) are measured with human errors during the tests (and such errors could be easily corrected
induced zero-sequence currents in other wye side or in delta windings. during the tests if the persons know the expected values for Z0). For

⁎
Corresponding author at: Dpto. de Conversión y Transporte de Energía, Universidad Simón Bolívar, Apdo. Postal 89.000, Caracas 1080, Venezuela.
E-mail address: [email protected] (E. Sorrentino).

https://doi.org/10.1016/j.epsr.2018.08.020
Received 20 May 2018; Received in revised form 10 August 2018; Accepted 29 August 2018
0378-7796/ © 2018 Elsevier B.V. All rights reserved.
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

Table 1 The measured positive-sequence short-circuit impedances are: Z12,

Main data of analyzed transformers. Z13, Z23 (subscript 3 is for tertiary-delta). The values for the positive-
Unit Connection MVA kV MS sequence equivalent circuit are: Z1, Z2, Z3.
This article considers that tertiary can be the most internal winding
1 YNynd 15/15/5 45/16.05/10 N (T21) or tertiary can be the most external winding (21T). Codes T21
2 YNynd 15/15/5 45/16.05/10 N
and 21T indicate the order of windings, from the innermost one to the
3 YNynd 25/25/8.33 45/16.05/10 N
4 YNynd 25/25/8.33 66/21/10 N
outermost one.
5 YNynd 25/25/8.33 45/16.05/10 N In case of YNyn transformers, there are only four possible mea-
6 YNynd 30/30/10 136/21/10 ? surements (Z1-O, Z2-O, Z1-S, Z2-S). That is, the nomenclature is equivalent
7 YNynd 30/30/10 132/16.05/10 Y to the YNynd case with open-delta.
8 YNynd 75/75/25 220/71/10 Y
All these possible measurements for Z0 are not available in the
9 YNynd 100/100/60 220/60/10 ?
10 YNynd 120/120/40 230/46/10 ? analyzed transformers because all these measurements are not man-
11 YNynd 120/120/40 230/45/10 Y datory in the current standards [11–13]. Furthermore, the measure-
12 YNynd 150/150/50 230/71/20 N ments of active power during these tests are not mandatory [11–13],
13 YNyn 5 34.5/13.8 N
and the number of units with available data about the angle of Z0 is
14 YNyn 40 115/12 Y
15 YNyn 100 259.05/55 B
very low.
Information about the presence of magnetic shunts on tank walls is
Note: Column “MS” indicates the presence of magnetic shunts on tank walls; “Y” not available for units 6, 9 and 10. This fact is not important for the
is Yes, “N” is No, “?” is unknown, and “B” indicates that measurements for both purpose of this article, since only Z0SC values are analyzed for units 6, 9
cases are available. These magnetic shunts on inner tank walls are included in and 10, and the presence of magnetic shunts on tank walls has no in-
some power transformers to reduce the tank heating due to leakage flux (a fluence on the developed rules for Z0SC.
detailed description about them can be found in Ref. [10]).

3. Zero-sequence magnetizing impedances

example, during the analysis of the test data for this research, different
mistakes were detected and corrected (in the reported ranges of mea- 3.1. Analysis of magnetic circuits
surement devices, in the reported taps of instrument transformers, and
in the use of per-unit system), and other mistakes were corrected during Fig. 1 shows the two main paths for the zero-sequence magnetic flux
some factory tests. These facts illustrate the practical usefulness of (ZSMF) during the tests for the measurement of zero-sequence mag-
having rules for estimating the Z0 values in order to reduce the prob- netizing impedances (Z1-O, Z2-O). In case of YNynd transformers, Fig. 1
ability of human errors during these tests. only shows the two wye windings because the delta is opened (this
The aim of this paper is the presentation of simple rules to estimate condition is similar to the YNyn connection).
the measurements of Z0 for 3-phase core-type transformers, from the ZSMF inside the winding connected to the source can return through
positive-sequence impedances. A conceptual description about the the tank, or through the space between this winding and the tank. The
nature of Z0 in these transformers is included because the proposed path through the tank has high magnetic permeability but there is a gap
rules are based on it. The proposed rules can be considered as a com- for the ZSMF (Z0 is “reactor-type”). On the other hand, the currents
plement for the rules of the aforementioned IEC standard [11]. The induced in the tank limit the ZSMF through the tank.
rules were verified with measurements on transformers from 5 to The ZSMF through the space between the tank and the winding
150 MVA, which are typically installed in substations of electric uti- connected to the source is not negligible, and this non-ferromagnetic
lities. area is different if this winding is the inner one or the outer one. Due to
this reason, the measured Z0 values are different when the inner
2. Main data of analyzed transformers and nomenclature for zero- winding or the outer winding is connected to the source. In general, Z2-
sequence impedances O is greater than Z1-O, and this fact has been previously demonstrated
with the help of a simplified magnetic circuit [21].
Table 1 shows the main data of 15 analyzed transformers, from 8 An oversimplification of the magnetic circuit would lead to obtain
different manufacturers. Data of unit 15 were taken from Ref. [1]. Only that the difference (ΔZ0M) between Z2-O and Z1-O is the positive-se-
the highest MVA rated value of YNynd transformers is identified in the quence impedance between these windings (Z12), and it has been pre-
following sections (and this value is considered to show all the im- viously shown [21] that this rule should not be applied in case of
pedances in per-unit).
Z0 of YNynd transformers can be measured by feeding a wye side
with the other wye side open-circuited or short-circuited, and the ter-
tiary delta can be connected or opened. This fact implies eight possible
Z0 measurements. The analysis of YNynd transformers requires the
study of YNyn and Dyn connections because one side of the YNynd
transformer can be without current during the test (and only the
windings with currents determine the magnetic fluxes within the
transformer). Therefore: (a) the YNynd transformer with open-delta is
similar to the YNyn connection; (b) the YNynd transformer with closed-
delta and with the other wye side open-circuited is similar to the Dyn
connection.
The eight possible Z0 measurements for the YNynd connection are:
Z1-O, Z2-O, Z1-S, Z2-S, Z1-O-D, Z2-O-D, Z1-S-D, Z2-S-D. The first subscript re-
presents the wye winding connected to the source during the test (1:
outer; 2: inner), the second subscript represents the connection of the
other wye winding (O: open; S: short-circuited), and the third subscript Fig. 1. Schematic view of main paths for ZSMF (arrows) during the measure-
represents the connection of the tertiary (none: open-delta; D: closed- ments of zero-sequence magnetizing impedances. Solid lines are related to
delta). fluxes through the tank. (a) Test for Z2-O; (b) test for Z1-O.

95
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

the same current in per-unit (if both values are taken at the same zero-
sequence voltage in per-unit, the approximation error is slightly greater
[21]). Results of Fig. 2 are useful to confirm that ΔZ0M tends to be
approximately constant if the difference is taken at the same current in
per-unit or at the same zero-sequence voltage in per-unit (i.e., obtained
curves are practically parallel).
On the other hand, the second rule related to Z0M is: the expected
values of Z1-O and Z2-O should be greater for units with magnetic shunts
on tanks walls than for units without magnetic shunts on tanks walls.
This rule is based on the fact that magnetic shunts on tank walls are
ferromagnetic paths which facilitate the ZSMF by the space between the
winding connected to the source and the tank (i.e., the aforementioned
values of Z0M are greater since the net reluctances are lower).
Furthermore, the use of magnetic shunts on tanks walls decreases the
ZSMF by the tank; consequently, tank losses during ZOM tests in theses
cases should be lower than in cases of units without magnetic shunts on
tank walls. Therefore, the third rule related to Z0M is: the expected
angles of Z0M should be greater for units with magnetic shunts on tanks
walls than for units without magnetic shunts on tanks walls (since va-
lues of Z0M are greater, and tank losses during ZOM tests are lower, in
cases of units with magnetic shunts on tanks walls).

3.2. Verification of proposed rules about the magnitudes of Z0M

Table 2 shows the available Z0M data for the analyzed transformers.
A subset of these data was analyzed in Ref. [21] to show that ΔZ0M is
approximately Z12 for units without magnetic shunts on tank walls but
not for units with magnetic shunts on tank walls (i.e., this fact is con-
Fig. 2. Example of nonlinearity of zero-sequence magnetizing impedances in firmed here). Two additional points can be highlighted from Table 2:
core-type transformers (the markers in the graph are the measured values in (a) Z1-0 is approximately in the range of 50%–90% for units without
unit 5).
magnetic shunts on tank walls, and in the range of 110%–150% for
units with magnetic shunts on tank walls; (b) K0 ≈ 1 for units without
transformers with magnetic shunts on tank walls. However, ΔZ0M is magnetic shunts on tank walls, and K0 is approximately in the range of
approximately Z12 for units without magnetic shunts on tank walls 0.55–0.75 for units with magnetic shunts on tank walls. Both points are
[21], and this fact is taken to formulate the first rule related to Z0M. On in accordance with the rules derived from the conceptual analysis
the other hand, for cases of transformers with magnetic shunts on tank shown in Section 3.1. These numerical ranges of values can be im-
walls, an additional analysis of measured and computed values of Z0M proved in the future, from the analysis of additional test data from other
in Ref. [21] indicates that ΔZ0M is lower than Z12 for this type of core-type transformers (especially in case of transformers with mag-
transformers; this point had not been previously highlighted and it is netic shunts on tank walls, because the number of analyzed units is very
also taken to formulate the first rule related to Z0M. Thus, the first rule low in this case, and the data of Z0M for unit 14 were not taken exactly
related to Z0M is: at the same per-unit value of IN).
ΔZ0M = Z2-O − Z1-O ≈ K0 Z12 (1)
3.3. Angle of Z0M
K0 = 1 for units without magnetic shunts on tank walls, otherwise
K0 < 1. Table 3 shows the available data for analyzing the angle of Z0M (and
Magnetizing Z0 values are dependent of nonlinearity of tank steel the resistive part of Z0M). A recommendation about using an estimated
[2–5,10,22]. An example of this nonlinearity is shown in Fig. 2. value near to 69° for the angle of Z0M is mentioned in Ref. [22], for
Therefore, the application of this rule should be performed under some transformers without magnetic shunts on tank walls. This angle is close
specific conditions. The recommendation from Ref. [21] is the appli- to 72.5°, which is an approximate value mentioned in Ref. [4] (without
cation of this approximation when both values (Z1-O, Z2-O) are taken at taking into account the presence or not of magnetic shunts). Results

Table 2
Available data of Z0M for the transformers taken as examples, and computed quotients (ΔZ0M/Z12).
Unit MVA Z1-O (%) Z2-O (%) ΔZ0M (%) Z12 (%) ΔZ0M/Z12 (pu) IN or V0 (pu) MS

13 5 50.20 56.99 6.79 6.92 0.98 V0 = 0.025 No

1 15 90.30 100.27 9.97 10.91 0.91 IN = 0.2 No
88.25 100.27 12.02 10.91 1.10 V0 = 0.07 No
3 25 75.05 86.19 11.14 10.76 1.04 IN = 0.3 No
70.62 83.32 12.70 10.76 1.18 V0 = 0.11 No
5 25 77.75 87.73 9.98 – – IN = 0.23 No
80.82 92.02 11.20 – – V0 = 0.04 No
14 40 152.88 – 10.42 13.90 0.75 IN = 0.17 Yes
– 163.30 IN = 0.12 Yes
15 100 109.0 116.0 7.0 12.80 0.55 IN = 0.3 Yes
55.40 69.00 13.60 12.80 1.06 IN = 0.3 No
11 120 112.8 – – – – IN = 0.16 Yes

96
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

Table 3 Table 4
Data of transformers to analyze the resistive part of Z0M. Estimation for Z1-0 using K1 values from Ref. [8].
Unit 1 3 14 Unit MVA MS K1 Z12 (%) Z1-0 (%) Z1-0 (%)
MVA 15 25 40 Estimated Measureda

Measured values Z12 10.91/87.15° 10.76/ 13.90/88.86° 13 5 N 11 6.92 76.12 50.20

(%) 87.48° 1 15 N 9.0 10.91 98.19 90.30
Z13 18.25/86.41° 20.03/ – 3 25 N 7.8 10.76 83.93 75.05
86.58° 5 25 N 7.8 10.68 83.30 77.75
Z23 5.26/79.43° 6.34/80.90° – 14 40 Y 6.7 13.90 93.13 152.88
Z1-O 88.25/65.59° 70.62/ 152.88/77.95° 15 100 N 4.6 12.80 58.88 55.40
70.18° 15 100 Y 4.6 12.80 58.88 109.00
Z2-O 100.27/ 83.32/ 163.30/77.78° 11 120 Y 4.3 13.78 59.25 112.80
67.71° 72.30°
a
Computed values R1-0 (%) 36.47 23.93 31.91 Measurement conditions are described in Table 2 for each specific case.
R2-0 (%) 38.03 25.33 34.57
3.5. Summary of proposed rules about Z0M
Note: R is the real part of the impedances.
Underlined values are the angles of complex numbers in their polar form.
First of all, it should be emphasized that Z0M values are nonlinear.
Section 3.2 shows that Z1-0 is approximately in the range of 50%–90%
from Table 3 show that these approximate values should not be applied
for units without magnetic shunts on tank walls, and in the range of
to units with magnetic shunts on tank walls. As mentioned in Section
110%–150% for units with magnetic shunts on tank walls. Eq. (1)
3.1, tank losses during ZOM test should be lower in this case (and,
shows that Z2-O can be estimated as Z1-O + K0 Z12 (Z2-O and Z1-O values
consequently, the angle of Z0M is higher). Thus, a coarse approximation
must be taken at the same current in per-unit or at the same zero-se-
is: an angle of Z0M near to 70° could be expected for units without
quence voltage in per-unit), and Section 3.2 shows that K0 ≈ 1 for units
magnetic shunts, and an angle of Z0M near to 80° could be expected for
without magnetic shunts on tank walls, and K0 is approximately in the
units with magnetic shunts on tank walls. Obviously, these rough rules
range of 0.55–0.75 for units with magnetic shunts on tank walls. Thus,
can be refined with more detail, in the future, by analyzing more Z0M
these rules allow the estimation of Z1-O and Z2-O values.
data.
On the other hand, Section 3.3 shows that a coarse approximation
for the angle of Z0M is: (a) near to 70° for units without magnetic shunts;
(b) near to 80° for units with magnetic shunts on tank walls.
3.4. Other rules from literature

Some authors have indicated that Z1-O could be estimated by mul- 4. Zero-sequence short-circuit impedances
tiplying Z12 by an empirical factor (K1) [7–9]:
4.1. Analysis of magnetic circuits
Z1-O ≈ K1 Z12 (2)

Proposed value of K1 [8,9] decreases as transformer capacity is The equilibrium of magnetomotive forces during Z0SC tests is dif-
larger (e.g., K1 ≈ 10 for 10 MVA, and K1 ≈ 5 for 80 MVA, as shown in ferent than in case of positive-sequence short-circuit tests. For positive-
Fig. 3). Table 4 shows the result of applying this estimation to the sequence short-circuit tests, N1I1 = N2I2 if there are currents only in
transformers taken as examples. This comparison shows that this rule two windings, and N1I1 = N2I2 + N3I3 if there are currents in the three
should not be applied for transformers with magnetic shunts on the windings (N1, N2, N3 are the numbers of turns; I1, I2, I3 are the winding
tank walls. This rule of thumb can be applied for units without mag- currents). Nevertheless, these equations are invalid for zero-sequence
netic shunts on tank walls, considering that differences between mea- tests since zero-sequence currents can circulate in the tank.
sured and estimated values can be in the order of 20%. a) Zero-sequence currents in only two windings
On the other hand, Table 1 of IEC Std. 60076-8 [11] simply in- In these cases, the zero-sequence magnetic fluxes (ZSMF) pass
dicates two “approximate” values for Z0M (Z2-0 ≈ 60% and Z1- through the zone between the winding connected to the source and the
0 ≈ 50%), which are not very near to the values shown in Table 2. short-circuited winding (the closed-delta is considered a short-circuited
Fortunately, Section 4.3 of this standard properly indicates that these winding for zero-sequence currents). This fact establishes the following
impedances are nonlinear. Measured values of Table 2 show that the similarities with positive-sequence impedance values: Z1-S and Z2-S are
mean value of Z1-0 is nearer to 70% than to 50%, for units without similar to Z12, Z2-O-D is similar to Z23, and Z1-O-D is similar to Z13.
magnetic shunts on tank walls, and the mean value of Z1-0 is near to Main return paths for ZSMF are imposed by the relative position of
130% for units with magnetic shunts on tank walls (variations in results short-circuited winding since the net flux linkages should be almost
are in the order of 20% in both cases). zero inside it (voltage drop in winding resistance can be considered
negligible). The main ZSMF paths for the different tests are shown in
Figs. 4 and 5. There are main return paths through the magnetic core if
the short-circuited winding is outer than the winding connected to the
power source (Figs. 4 a, 5 a, c, and d), otherwise there are main return
paths through the tank (Figs. 4 b–d, and 5 b).
When the main return paths are through the core, the Z0SC value is
very similar to the sequence-positive value:

Z2-S ≈ Z12 (3)

Z2-O-D ≈ Z23 for case 21T (4)

Z1-O-D ≈ Z13 for case 21T (5)

When the main return paths are through the tank, the Z0SC value is
Fig. 3. Values of K1 (figure reproduced from Ref. [8]). lower than the sequence-positive value (and the stray losses are greater

97
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

Fig. 4. Schematic view of main paths for ZSMF during Z0SC tests, with zero-sequence currents in only two windings. Case: T21.

because there are more losses in the tank):

Z1-S < Z12 (6)

Z2-O-D < Z23 for case T21 (7)

Z1-O-D < Z13 for case T21 (8)

b) Zero-sequence currents in the three windings

In these cases, one winding is connected to the power source and
two windings are short-circuited. The reference values for the Z0SC
values will be the equivalent positive-sequence impedances for the
same condition:
Fig. 6. Schematic view of main paths for ZSMF during Z0SC tests, with zero-
Z123 ≈ Z1 + Z2Z3/(Z2 + Z3) (9) sequence currents in the three windings. Case: T21.

Z213 ≈ Z2 + Z1Z3/(Z1 + Z3) (10)

Z123 is the sequence-positive equivalent impedance when winding 1

is connected to the source and the other windings are short-circuited.
Z213 is the correspondent value when winding 2 is connected to the
source.
Figs. 6 and 7 show the ZSMF paths for the different tests, for cases
T21 and 21T, respectively. There are main return paths of ZSMF
through the tank only for Z1-S-D in case T21 (Fig. 6b). Thus:

Z2-S-D ≈ Z213 (11)

Z1-S-D ≈ Z123 for case 21T (12)

Z1-S-D < Z123 for case T21 (13) Fig. 7. Schematic view of main paths for ZSMF during Z0SC tests, with zero-
sequence currents in the three windings. Case: 21T.

cases of the inequalities:

4.2. Verification of proposed rules
a) Zero-sequence currents in only two windings

Table 5 shows the data for the 15 transformers taken as examples, as Z1-S ≈ 0.91 Z12 (14)
well as computed quotients of Z0SC divided by the correspondent po-
Z2-O-D ≈ 0.95 Z23 for case T21 (15)
sitive-sequence impedances. These quotients show that approximations
from Eqs. (3) to (13) are very good. Z1-O-D ≈ 0.87 Z13 for case T21 (16)
The computed quotients allow the following refinement for the

Fig. 5. Schematic view of main paths for ZSMF during Z0SC tests, with zero-sequence currents in only two windings. Case: 21T.

98
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

Table 5
Data of Z0SC for fifteen transformers taken as examples, and computed quotients for Z0SC.
Connection YNynd (T21) YNynd (21T) YNyn

Unit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
MVA 15 15 25 25 25 30 30 75 100 120 120 150 5 40 100

Z12 10.91 10.64 10.76 10.90 10.68 10.90 14.06 14.00 8.87 14.45 13.78 14.11 6.92 13.90 12.80
Z13 18.25 17.55 20.03 19.11 18.10 17.10 21.39 20.79 14.58 21.69 22.47 17.65 – – –
Z23 5.26 4.98 6.34 5.55 4.75 4.26 5.04 4.92 4.58 40.17 40.53 35.57 – – –

Z1-O-D 16.28 – 16.47 – 15.17 – – 19.40 13.60 20.88 21.91 – – – –

Z1-S-D 9.95 8.81 9.45 9.59 9.20 10.10 12.87 13.30 8.49 7.65 7.46 7.01 – – –
Z2-O-D 5.29 – 5.66 – 4.72 – – 4.81 4.46 38.86 – – – – –
Z2-S-D 3.39 3.20 3.3 3.12 3.03 2.67 3.28 3.32 – 14.39 – 14.04 – – –
Z1-S 10.17 9.13 9.73 9.85 9.56 10.30 13.12 13.40 – 13.07 12.95 12.71 6.20 13.24 12.10
Z2-S – 10.41 10.77 11.00 11.00 11.00 14.06 13.60 – 14.57 – 14.16 6.83 – –

Z2-S/Z12 – 0.98 1.00 1.01 1.03 1.01 1.00 0.97 – 1.01 – 1.00 0.99 – –
Z1-S/Z12 0.93 0.86 0.90 0.90 0.90 0.94 0.93 0.96 – 0.90 0.94 0.90 0.90 0.95 0.95
Z2-O-D/Z23 1.01 – 0.89 – 0.99 – – 0.98 0.97 0.97 – – – – –
Z1-O-D/Z13 0.89 – 0.82 – 0.84 – – 0.93 0.93 0.96 0.98 – – – –
Z2-S-D/Z213 1.10 1.08 1.00 1.02 1.10 1.00 1.01 1.01 – 1.01 – 1.01 – – –
Z1-S-D/Z123 0.93 0.84 0.91 0.91 0.89 0.95 0.93 0.96 0.96 0.99 0.99 1.02 – – –

b) Zero-sequence currents in the three windings Table 7

Application of these Z0SC rules to an example from the literature.
Z1-S-D ≈ 0.9 Z123 for case T21 (17)
Value Estimated Measured [24,25] Error in the quotient
These four factors (0.91, 0.95, 0.87, 0.9) were selected in order to
minimize the maximum errors in comparison with the quotients shown Z1-S 0.91 Z12 = 12.54% 12.82% 0.02
Z2-S Z12 = 13.78% 13.63% −0.01
in Table 5. The other factors, in Eqs. (3)–(5) and (9)–(12), are equal to
Z1-O-D 0.87 Z13 = 28.97% 28.89% 0.03
unity. Table 6 shows the errors between these factors and the quotients Z2-O-D 0.95 Z23 = 17.10% 17.21% 0.01
computed from the measured data. Each error is the difference of the Z1-S-D 0.9 Z123 = 12.37% 12.82% 0.03
computed quotients minus the factors in Eqs. (3)–(5), (9)–(12), and Z2-S-D Z213 = 7.43% 7.36% −0.01
(14)–(17). A wider set of transformers can be evaluated, of course, in
order to improve these factors and the certainty about their accuracy.
Almost all the errors in Table 5 are lower than 0.07. The three ex- the tank. Therefore, the rules are:
ceptions are for Z2-S-D in case T21, and their errors are also acceptably a) Zero-sequence currents in only two windings
low. For these three exceptions, the measured value is slightly higher Figs. 4 and 5 show when the main paths for ZSMF are through the
than the expected value. In other similar cases, the detected mistake tank (for these cases, the effective resistance must be greater than in
was an excessive voltage drop in the conductors (thus, the measured positive-sequence). Thus:
voltage is slightly higher than the voltage at transformer terminals); R2-S ≈ R12 (18)
however, this mistake only can be corrected during the tests.
R2-O-D ≈ R23 for case 21T (19)
4.3. Application of the proposed rules to an example from the literature R1-O-D ≈ R13 for case 21T (20)

These rules were also applied to an example (T21) from the litera- R1-S > R12 (21)
ture. The measured positive-sequence impedances are [24,25]: R2-O-D > R23 for case T21 (22)
Z12 = 13.78%, Z13 = 33.3%, Z23 = 18%. Table 7 shows the compar-
ison between estimated and measured values. These results also con- R1-O-D > R13 for case T21 (23)
firm the validity of these rules.
b) Zero-sequence currents in the three windings
Figs. 6 and 7 show the main path for ZSMF. Thus:
4.4. Resistive part of Z0SC
R2-S-D ≈ R213 (24)
Section 4.1 indicates that stray losses are greater when main return
R1-S-D ≈ R123 for case 21T (25)
paths for ZSMF are through the tank because there are more losses in

Table 6
Errors in computed quotients for Z0SC, in comparison with the factors of Eqs. (3), (7)–(12), and (14)–(17).
Connection YNynd (T21) YNynd (21T) YNyn

Unit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
MVA 15 15 25 25 25 30 30 75 100 120 120 150 5 40 100

Z2-S/Z12 – −0.02 0.00 0.01 0.03 0.01 0.00 −0.03 – 0.01 – – −0.01 – –
Z1-S/Z12 0.02 −0.05 −0.01 −0.01 −0.01 0.03 0.02 0.05 – −0.01 0.03 0.03 −0.01 0.04 0.04
Z2-O-D/Z23 0.06 – −0.06 – 0.04 – 0.03 0.02 −0.03 – – – – –
Z1-O-D/Z13 0.02 – −0.05 – −0.03 – 0.04 0.04 −0.04 −0.02 −0.02 – – –
Z2-S-D/Z213 0.10 0.08 0.00 0.02 0.10 0.00 0.01 0.01 – 0.01 – – – – –
Z1-S-D/Z123 0.03 −0.06 0.01 0.01 −0.01 0.05 0.03 0.06 0.06 −0.01 −0.01 −0.01 – – –

99
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

Table 8
Data of transformers for analyzing the resistive part of Z0SC.
Unit 1 3 9 14
Connection YNynd (T21) YNynd (T21) YNynd (T21) YNyn
MVA 15 25 100 40

Measured values (%) Z12 10.91/87.15° 10.76/87.48° 8.87/88.78° 13.90/88.86°

Z13 18.25/86.41° 20.03/86.58° 14.58/88.95° –
Z23 5.26/79.43° 6.34/80.90° 4.58/87.47° –
Z1-O-D 16.28/82.81° 16.47/82.54° 13.60/88.59° –
Z1-S-D 9.95/83.81° 9.45/84.08° 8.49/88.25° –
Z2-O-D 5.29/78.17° 5.66/79.38° 4.46/87.11° –
Z2-S-D 3.39/79.04° 3.3/81.37° – –
Z1-S 10.17/84.51° 9.73/84.72° – 13.24/87.92°
Z2-S – 10.77/87.48° – –
Computed values (%) R2-S/R12 – 1.00 – –
R1-S/R12 1.79 1.89 – 1.74
R2-O-D/R23 1.12 1.04 1.11 –
R1-O-D/R13 1.78 1.79 1.25 –
R2-S-D/R213 1.15 0.96 – –
R1-S-D/R123 1.64 1.64 1.26 –

Note: R is the real part of impedances.

Underlined values are the angles of complex numbers in their polar form.

0.8 < a1 < a2 < a3 < 1 (24)

R1-S-D > R123 for case T21 (26)
Therefore, the rules of this paper are in accordance with the IEC Std.
Table 8 shows the available data and computed quotients of re-
60076-8 and can be considered as a complement for the standard.
sistive part of Z0SC divided by the resistive part of the correspondent
Values of a2 and a3 are near to unity, and the value of a1 has been
positive-sequence impedances. These quotients confirm the proposed
specified for each case. The accuracy of this refinement for a1 could be
rules for cases T21 and YNyn (cases 21T could not be confirmed, due to
improved in the future, by including data from a wider set of trans-
unavailability of data).
formers.
R2-O-D/R23 is lower than R1-S/R12 and R1-O-D/R13 because the case of
On the other hand, the Eqs. (18)–(26) summarize the proposed rules
Z2-O-D has longer distance between tank and winding connected to the
for the resistive part of Z0SC. IEC Std. 60076-8 does not give information
source. Due to the same reason, the proposed factor in Section 4.2 for
about these values.
Z2-O-D (0.95) is the highest one.
Units 9 and 14 have magnetic shunts on tank walls. Due to this fact,
the quotients which are larger than unity should be lower for these 5. Conclusion
cases. This result is obtained for R1-S-D/R123 and R1-O-D/R13 in unit 9
(and not for R2-O-D/R23, because this quotient is very near to unity, due Simple rules to estimate the results of measurements of zero-se-
to the reason explained in the previous paragraph). For unit 14, R1-S/ quence impedances for 3-phase core-type transformers were developed.
R12 is lower than for units without magnetic shunts on tank walls but In case of magnetizing zero-sequence impedances, the presence or not
the difference shown in Table 8 is very low (and greater differences can of magnetic shunts on tank walls defines the details of the proposed
be expected in other cases). Again, more refinement of these rules can rules. For short-circuit zero-sequence impedances, only the positive-
be obtained in the future by analyzing more data. sequence transformer impedances are necessary to apply the proposed
rules. These rules were applied to a set of fifteen transformers, from 5 to
150 MVA. The proposed rules were conceptually developed from the
4.5. Summary of proposed rules about Z0SC and comparison with IEC
analysis of magnetic circuits of these transformers, and the data from
standard 60076-8
standardized tests were important in order to obtain a numerical re-
finement for the estimation of the measurements of zero-sequence im-
Table 9 shows a summary of the proposed rules about Z0SC, taken
pedances. This numerical refinement was developed for transformers
from Eqs. (3) to (17), as well as a comparison between these rules and
which are typically installed in the substations of electric utilities. A
the rules from the IEC Std. 60076-8 [11]. This standard utilizes three
wider set of transformers can be evaluated in the future, in order to
factors (a1, a2, and a3), and it only gives the following relationship for
improve the accuracy of such refinement.
them:
Some rules for the real part (or the angle) of zero-sequence im-
pedances were also developed, and they were also verified with test
Table 9 data. However, there are few available data for this purpose, and a
Comparison of Z0SC rules of this paper with IEC Std. 60076-8. refinement similar to the obtained one for the module of zero-sequence
Case Rules of this paper IEC Std. 60076-8 impedances was not possible.
The practical usefulness of developed rules is related to the reduc-
Both Z1-S ≈ 0.91 Z12 Z1-S ≈ a1 Z12 tion of probability of human errors during the tests, as well as with the
Both Z2-S ≈ Z12 Z2-S ≈ a2 Z12
analysis of test results. For example, during the analysis of test data for
T21 Z2-O-D ≈ 0.95 Z23 Z2-O-D ≈ a1 Z23
Z1-O-D ≈ 0.87 Z13 Z1-O-D ≈ a1 Z13 this research, some mistakes were corrected (but some mistakes only
21T Z2-O-D ≈ Z23 Z2-O-D ≈ a2 Z23 can be corrected during the tests).
Z1-O-D ≈ Z13 Z1-O-D ≈ a2 Z13 The developed rules can be seen as a complement for the rules of the
T21 Z1-S-D ≈ 0.9 Z123 Z1-S-D ≈ a1 Z123 IEC Std. 60076-8 because: (a) the physical fundamentals for the rules
21T Z1-S-D ≈ Z123 Z1-S-D ≈ a2 Z123
have been explained here; (b) some numerical factors, from real tests,
T21 Z2-S-D ≈ Z213 Z2-S-D ≈ a2 Z213
21T Z2-S-D ≈ a3 Z213 have been included here, and these factors can be updated in the future
by using more test data of transformers; (c) some new rules, for the real

100
E. Sorrentino, J.C. Burgos Electric Power Systems Research 165 (2018) 94–101

part or the angle of zero-sequence impedances, were included here; (d) [10] S. Kulkarni, S. Khaparde, Transformer Engineering: Design and Practice, Marcel
a rule to estimate the difference between the two magnetizing im- Dekker Inc., 2004.
[11] IEC Std. 60076-8, Power Transformers. Part 8: Application Guide, 1997.
pedances is included here; (e) the presence of magnetic shunts on tank [12] IEC Std. 60076-1, Power Transformers. Part 1: General, 2011.
walls has an effect on results and this effect has been considered here. [13] IEEE Std. C57.12.90, IEEE Standard Test Code for Liquid-immersed Distribution,
Power, and Regulating Transformers, 2010.
[14] R. Del Vecchio, B. Poulin, P. Feghali, D. Shah, R. Ahuja, Transformer Design
Acknowledgements Principles, CRC Press, 2010.
[15] M. Heathcote, The J&P Transformer Book, Reed Educational and Professional
Authors are grateful to Gas Natural Fenosa (Spain), Siemens Publishing Ltd., 1998.
[16] C. Wagner, R. Evans, Symmetrical Components, McGraw-Hill, 1933.
(Colombia and Venezuela) and CAIVET (Venezuela) for their valuable [17] J. Winders, Power Transformers: Principles and Applications, Marcel Dekker, 2002.
help, by providing the test data of transformers. [18] R. Bean, N. Chackan, H. Moore, E. Wentz, Transformers for the Electric Power
Industry, McGraw-Hill, 1959.
[19] A. Blume, G. Boyajian, T. Camilli, S. Lennox, V. Minneci, Transformers Engineering,
References
John Wiley & Sons, 1951.
[20] Westinghouse Electric Corporation, Electrical Transmission and Distribution
[1] A. Carlson, J. Fuhr, G. Schemel, F. Wegscheider, Testing of Power Transformers, Reference Book, Westinghouse Electric Corporation, 1950.
ABB AG, 2003. [21] E. Sorrentino, J. Burgos, About the difference between zero-sequence magnetizing
[2] P. Anderson, Analysis of Faulted Power Systems, IEEE Press, 1995. impedances of a 3-phase core-type transformer, Electr. Power Syst. Res. 116
[3] E. Clarke, Circuit Analysis of AC Power Systems vol. 2, John Wiley & Sons, 1950. (November) (2014) 439–443.
[4] A. Garin, Zero-phase-sequence characteristics of transformers. Part II, Gen. Electr. [22] E. Sorrentino, J. Burgos, Application of 2D linear models for computing zero-se-
Rev. 43 (April (4)) (1940) 174–179. quence magnetizing impedances of 3-phase core-type transformers, Electr. Power
[5] A. Garin, Zero-phase-sequence characteristics of transformers. Part I, Gen. Electr. Syst. Res. 117 (December) (2014) 123–133.
Rev. 43 (March (3)) (1940) 131–136. [23] E. Sorrentino, J. Burgos, Application of 2D linear modeling for computing zero-
[6] M. Christoffel, Nullreaktanzen von transformatoren und drosselspulen, Brown sequence short-circuit impedances of 3-phase core-type YNynd transformers, Electr.
Bovery Mitteilungen 52 (November/December (11/12)) (1965) 837–842 (in Power Syst. Res. 122 (2015) 1–9.
German). [24] A. Narang, R. Brierley, Topology based magnetic model for steady-state and tran-
[7] C. Arturi, Comportamento omopolare dei trasformatori trifasi con nucleo a tre co- sient studies for three-phase core type transformers, IEEE Trans. Power Deliv. 9
lonne e due avvolgimenti, L’Energia Elettrica (1) (1987) (in Italian). (August (3)) (1994) 1337–1349.
[8] F. Coppadoro, L’impedenza alla sequenza zero dei trasformatori, Elettrificazione (2) [25] X. Chen, S. Venkata, A three-phase three-winding core-type transformer model for
(1970) (in Italian). low-frequency transient studies, IEEE Trans. Power Deliv. 12 (April (2)) (1997)
[9] V. Oels, Ersatszchaltungen des transformators für das nullsystem, Elektrotechnische 775–782.
Zeitschrift Ausgabe 89 (1968) 59–62 (in German).

101