See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/256354678

Five-Leg transformer model for GIC studies

Conference Paper · July 2013

CITATIONS READS

9 398

6 authors, including:

Nicola Chiesa Abbas Lotfi

Statoil ASA Nexans Norway AS
37 PUBLICATIONS   495 CITATIONS    16 PUBLICATIONS   83 CITATIONS   

SEE PROFILE SEE PROFILE

Hans Kristian Høidalen

Norwegian University of Science and Technology
114 PUBLICATIONS   1,473 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Submarine HVDC cable system View project

Zero sequence impedance modeling of transformers View project

All content following this page was uploaded by Abbas Lotfi on 19 May 2014.

The user has requested enhancement of the downloaded file.

 Five-leg transformer model for GIC studies
Nicola Chiesa, Abbas Lotfi, Hans K. Høidalen, Bruce Mork, Øyvind Rui, Trond Ohnstad

Abstract—Geomagnetic induced currents (GIC) may saturate nonlinearity of ferromagnetic materials, the coupling and
transformer cores and lead to increased reactive power connection of coils, and the system series resistances.
consumption and disoperation of power systems. The paper The GIC phenomenon is analyzed in several papers for
analyses the influence of transformer core topology and the air- both single- and three-phase transformers. These include
path inductances on the GIC response. Air-paths inductances are
FEM-based models [6,7], and magnetic circuit models [8-14].
calculated with finite element method and are shown to be much
larger for DC than for 50 Hz. Air-path inductances play a The FEM models can be very accurate particularly for the
significant role for reactive power consumption and differential cases where power losses and temperature rises in the tank and
currents due to the deep saturation of the outer limbs of the 5- other metallic components are of interest. However, the
legged transformer. The paper analyzes reactive power requirement of detailed design information and the
consumption and differential current harmonics related to power computational burden makes the application of FEM modeling
system protection. not viable for power system studies.
The models based on magnetic circuit theory have been
Keywords: Transformer Modeling, GIC, Duality Transfor- widely used for the analysis of GIC effects on transformers.
mation, Hybrid Model, Zero Sequence Flux, FEM
Some models consider only very basic magnetic circuits
lacking a topological representation of the core structure and
I. INTRODUCTION
DC flux paths, and using only simple calculation of model

G eomagnetically induced currents (GICs) are an effect of

solar storm and induced geomagnetic disturbances on the
power networks. Charged particles coming from the sun
parameters [8,9]. More advanced models proposed for GIC
studies are based on a topologically correct core representation
modeling independent core sections and air-flux paths [10-15].
interact with Earth’s magnetosphere-ionosphere and produce In particular, [2] analyses a 5-legged transformer topology and
ionosphere currents that perturb Earth’s geomagnetic field, saturation of core elements. However, two main issues are not
inducing an Earth’s surface potential gradients (3-6 V/km). considered in these models. The first issue is the role of air-
The voltage difference between two grounding points results path reluctances on the flux distribution in different flux paths.
in the flow of zero-sequence currents called GIC [1,2]. The The second one is the investigation of DC flux offset effect on
frequencies of these currents are very low and can be the amplitude and harmonic content of the magnetizing
considered as quasi-DC. The duration of these incidents can current, as well as on reactive power consumption.
be in the order of several minutes to several hours [1,3]. DC flux impressed in the windings is in the same direction
A DC current flowing through the windings of a power for all legs, therefore the flux will flow through the zero
transformers generates a DC flux in the core with a magnitude sequence flux (ZSF) path. In the case of three-phase three-leg
depending on the magnitude of the DC current, number of constructions, the ZSF goes outside the core, flows through
turns in the windings carrying the current and reluctance of the the air gap and tank and returns to the core. For five-leg
DC flux path. This impressed DC flux shifts the operating constructions, the lateral legs act as return paths for ZSF. As
point of the magnetizing characteristic and causes half cycle long as they do not saturate, the air-paths and the tank will
saturation in the core resulting in harmonic currents, forcing have no considerable effect. For high level of GIC the lateral
the flux to flow outside the core and increased reactive power legs saturate and part of the flux flows outside the core
consumption [4,5]. through the oil and the tank leading to increased losses and the
The important aspects to consider for analyzing a DC temperature rise in the tank.
biased power transformer are the topology of the core, the This paper contributes with a modeling approach for the
analysis of the effect of GIC in power transformers with focus
N. Chiesa is with Sintef Energy Research, Norway, (Nicola.Chiesa@sintef.no) on air-path inductances estimation and sensitivity, harmonic
A. Lotfi and H. Kr. Høidalen are with Norwegian University of Science and content of the magnetizing current, reactive power
Technology (NTNU), Electric Power Engineering Department, N-7034 consumption and GIC influence on protection relays.
Trondheim, Norway, (abbas.lotfi@ntnu.no).
B. Mork is with Michigan Technological University (MTU), Houghton, Section II describes the transformer test object and modeling
Michigan, USA (bamork@mtu.edu). of the associated source and loads, as well as the GIC
Ø. Rui and T. Ohnstad. are with Statnett, Trondheim, Norway excitation. Section III describes calculation of air-path
This work was supported in part by the KMB project " Electromagnetic
transients in future power systems" financed by the Research Council of
inductances based on a finite element approach. Section IV
Norway and by Dong Energy, EdF, EirGrid, Hafslund, National Grid, Nexans, shows calculation results of internal fluxes, power
RTE, Siemens WP, Statnett, Statkraft, and Vestas WS. consumption, and current harmonics. Section V discusses the
results with respect to relay protection and application of the
Paper submitted to the International Conference on Power Systems
Transients (IPST2013) in Vancouver, Canada July 18-20, 2013. modeling approach to three-limb transformers.
 𝑉𝐷𝐶
II. MODELING OF GIC IN TRANSFORMERS 𝐺𝐼𝐶 = (1)
Re�𝑍𝑔 �+𝑅𝑊
Fig. 1 shows the system topology considered for the analysis with RW the HV winding resistance. The reference to the GIC
of GIC. Table I gives the rating and design information of the amplitude instead of the DC voltage offset makes the analysis
300 MVA five-limb power transformer used in this study. The of GIC more convenient as the value of Zg does not influence
transformer has been modeled using the Hybrid transformer the results.
model approach [16,17]. No-load test report up to 115% Equation (1) is derived by the saturation equilibrium
excitation was available, as well as additional test and design condition [2]:
data useful for parameter estimations and for the FEM - a quasi-DC earth-surface voltage establish a DC core flux
modeling presented in Section III. The air-core inductance of offset;
HV winding is estimated to be 2.22 pu of leakage inductance - the flux offset drives the transformer into saturation leading
between high- and low-voltage winding. An accurate to the flow of unidirectional current pulses;
estimation of air-core inductance is essential for GIC studies - the DC component of current pulses generates a DC voltage
due to the high saturation regime. drop on the system series resistances;
- the flux offset continues to increase at a decreasing rate until
the voltage drop equals the DC earth-surface voltage
eliminating the DC voltage applied to the transformer;
- when this equilibrium is reached the half-cycle saturation
reaches a steady-state condition as long as the DC source is
present.
The saturation equilibrium is demonstrated in Fig. 3 and
Fig. 4 where the system of Fig. 1 is simulated with DC neutral
voltage applied and no AC source. These figures have a
Fig. 1: System topology. conceptual meaning only as superposition of the AC and DC
fluxes is not valid for a nonlinear system. The lack of air-paths
TABLE I
TRANSFORMER RATING AND DESIGN
and the homopolar flux in the transformer leg, impose a fixed
HV 300 kV 300 MVA YN flux ratio of 0.5 in the yoke and of 1.5 in the outer-limbs. A
LV 132 kV 300 MVA yn0 reduced cross-section area of the yokes and outer-limbs results
TV 47 kV 100 MVA d11 into a high flux density, leading first the outer-limbs in
AYOKE/ALEG=0.6 LYOKE/LLEG=2 saturation.
AOUTER/ALEG=0.6 LOUTER/LLEG=2.25

Fig. 2 shows the equivalent electrical circuit of a 5-legged

transformer core (excluding losses). This is similar to what is
used in the Hybrid Transformer model [16] except for the
additional air-path inductances Lair.

Fig. 3: Flux distribution in zero-sequence excitation without air-paths,

α β γ
with indication of relative flux densities.
Iα Iβ Iγ
λ Lair λ Lair λ Lair
1.2
Leg i4 Leg
i5 Leg i6 B[pu] Outer-limb
Out

Out
Yoke

Yoke

1.0
λo λy λy λo
λλ 0.8
Fig. 2: 5-legged transformer core electric topology.
0.6
The transformer is loaded at 73% with ZL = 100+jω10-4 Ω Leg
on secondary and ZT = 50 Ω on tertiary winding. Zg is a 0.4

negligible resistive series source impedance of 20 mΩ. Yoke

The representation of GIC in EMT studies is obtained by 0.2

imposing a DC voltage offset on the HV neutral point. To

avoid a step excitation and to resemble the longsome nature of 0.0
0 15 30 45 60 75 [s] 90
GIC, the DC voltage is applied with a slow ramp between 2 Fig. 4: Per unit magnetic field in different core sections with neutral
and 10 s. A direct injection of a DC current on the neutral voltage applied according to Fig. 1 without any air-path inductances
point is not suitable as this would not allow any flow of and AC source applied.
current harmonics. The value of the DC voltage offset and the
series resistance determine the amplitude of GIC: The effect of the air-path across each leg is considered in
Fig. 5 and Fig. 6. Flux starts flowing into air-paths only after maximum and minimum taps, respectively.
the outer-limb are fully saturated (in Fig. 6 after 46s). The
effect seen in Fig. 6 is that the flux density reached in the legs
at saturation equilibrium is higher than in the case without air-
paths. This results in different relative flux density distribution
in the core as indicated in Fig. 5.

Fig. 7: 3D-model of the case study.

TABLE II
CALCULATED AND MEASURED VALUES OF ZSI
Fig. 5: Flux distribution in zero-sequence excitation including air-
paths, with indication of relative flux densities at saturation Windings Meas- 3D- Error
Tap
equilibrium. Excited/shorted ured FEM %
HV300kV/TW47kV 13 107,2 105.18 1.88
1.2
HV300kV/TW47kV 1 116.8 115.56 1.06
B[pu] Outer-limb
HV300kV/TW47kV 25 99.69 97.9 1.8
1.0
LV132kV/TW47kV - 31.85 32.63 -2.45
0.8

When GIC occurs, there is a DC flux in each leg that can

Leg
0.6 cause the lateral legs to saturate. In this case the flux goes
Yoke outside the core and returns via air (oil) gaps and the tank as
0.4 shown in Fig. 8. The inductances corresponding to these paths
Air-path appear in the core model shown in Fig. 2 obtained by duality
0.2
transformation and is called Lair.

0.0
0 15 30 45 60 75 [s] 90
Fig. 6: Per unit magnetic field in different core sections with neutral
voltage applied according to Fig. 1 with air-path inductances added
(500 mH), but without AC source applied.

III. AIR PATH INDUCTANCE CALCULATION

In order to quantify air-path inductance values, 3D-FEM
calculations are performed for the transformer in Table I.
Fig. 7 shows this transformer modeled in ANSOFT Fig. 8: Flux Paths outside the core
MAXWELL. Design details for winding, core, and tank were
available but this cannot be brought here. Extended zero- The air paths inductances cannot be obtained from typical
sequence tests were also available. Because of symmetry, only zero-sequence tests, because the measured inductances are
half of the transformer is modeled as this dramatically associated with the combination of the lateral leg in parallel to
decreases solving time and memory requirements. The the relevant air-path. These cannot be separated from each
innermost winding of the transformer is the low voltage other for moderate excitations. In the case of FEM modeling,
winding (132 kV), the next are high voltage winding (300 the flux can be forced to go outside the core by placing two
kV), and tertiary winding (47 kV), respectively. fictitious windings on the lateral legs as shown in Fig. 8.
Running the FEM software in Eddy current and Magneto- Applying the same exciting current to the mentioned windings
static mode and using the energy method, the inductance and calculating the inductance matrix, the voltage of each
matrix is calculated. The short-circuit zero-sequence winding will be:
impedances (SCZSI) at different cases corresponding to the 𝑉𝑖 = 𝑗𝜔 ∑5𝑘=1 𝐿𝑘𝑖 𝐼𝑘 , 𝑖 = 1 … 5 (2)
tests done on this transformer are calculated, for the purpose The currents are the same, so:
of verifying the FEM model. 𝑉𝑖 = 𝐼𝑝 ∙ (𝑗𝜔 ∑5𝑘=1 𝐿𝑘𝑖 ), 𝐼𝑘 = 𝐼𝑝 , 𝑘 = 1 … 5 (3)
Table II shows the SCZSI obtained by 3D-FEM simulation This results in:
compared with the test results. The obtained values illustrate 𝑉
𝐿𝑎𝑖𝑟,𝑖 = 𝑖 → 𝐿𝑎𝑖𝑟,𝑖 = ∑5𝑘=1 𝐿𝑘𝑖 (4)
good agreement between the test results and FEM 𝐼𝑝

calculations. Tap numbers 13, 1 and 25 refer to principal, The air–paths shown in Fig. 8 also involve the tank. Thus
𝐿𝑎𝑖𝑟,𝑖 is the sum of the air-gap and tank inductance. Since the reactive power √𝑆 2 − 𝑃2 at saturation equilibrium; however,
penetration depth of magnetic flux in tank decreases with under distorted conditions this definition includes harmonic
frequency, the equivalent reluctance of tank increases leading reactive power components. Since only fundamental lagging
to a decrease of the corresponding inductance. For the current components have a significant impact on system
transformer modeled in this paper values of these inductances voltage profile [15], the fundamental reactive power is defined
are as shown in Table III. The inductances values are referred as:
to the innermost LV winding. As can be seen in table III, this 2
decrease in the inductance caused by magnetic penetration 𝑄50𝐻𝑧 = �3𝑉 2 𝐼50𝐻𝑧 − 𝑃2 (5)
depth is considerable even at 50 Hz excitation. with I50Hz being the fundamental harmonic of the current and
assuming that the voltage has no harmonic distortion.
TABLE III
AIR-PATH INDUCTANCES CALCULATED BY 3D-FEM
300
Inductance corresponding to the DC AC Excitation, 50
*10 6
air-path of Fig.8 (mH) Excitation Hz
250
1, 3 736.65 225.895 S [MVA]
2 771.99 242.593 200
P [MW]
The corresponding inductances are not voltage dependent 150
for expected amp-turns generated by GIC phenomenon since
the reluctance of the air-paths dramatically dominate the iron 100
sections (leg and the tank). In other words, it is not expected Q [MVAr]
that the tank walls, cover and bottom will be heavily saturated. 50

However, the magnetic penetration depth in the tank makes

the air-path inductances frequency dependent. If a detailed 0
0 15 30 45 60 75 [s] 90
investigation of the flux outside the core is of interest, the Fig. 9: S, P and Q variation due to GIC (100 A), with air-path
inductance Lair should be considered as a frequency dependent inductances (500 mH).
parameter. In this case the tank losses can be taken into
account for instance as a resistive element in combination with
inductive branches in a Cauer circuit. TABLE IV
For the purpose of this paper we consider an average, POWER AND CURRENT INTO TRANSFORMER
frequency-independent value for investigating the effect of the 100A GIC at Saturation Equilibrium
Steady
Without
Lair on terminal quantities. State 250 mH 500mH 750mH
air-path
S [MVA] 219 219 219 220 220
IV. RESULTS P [MW] 218 214 213 213 213
The system of Fig. 1 is simulated with different level of GIC Q [MVAr] 18 47 52 54 56
from 20 to 500 A and different values of air-path inductance. ΔQ [MVAr] - 29 34 36 38
The base reference case is for 100 A GIC and 500 mH/phase Q50Hz [MVAr] 8 36 39 40 41
air-path inductances. Irms [A] 419 421 422 423 424
Fig. 9 shows the increased reactive power consumption as a
well-known effect of GIC. For this specific transformer a 200
100 A GIC results in an increased 36 MVAr reactive power
demand. Since GIC is a system-wide phenomenon affecting 150
Q [MVAr]

several transformers simultaneously, power utilities are

100
particularly concerned about extraordinary reactive power
demand due to voltage stability concerns and reactive power 50
reserve scheduling.
Table IV compares the simulation results at steady-state (no 0
0 100 200 300 400 500
GIC) with the case of 100 A GIC for different values of air- GIC [A]
path inductance. While there is a minor effect of the GIC on Fig. 10: Reactive power consumption as function of GIC.
active power and terminal RMS current, the reactive power
increases by a factor of 3 for the analyzed system configura- Fig. 11 and Fig. 12 are the extension of Fig. 4 and Fig. 6,
tion. Not considering the air-path inductance has a respectively, with both AC source and neutral voltage applied
considerable effect on the increased reactive power demand according to Fig. 1. It is harder to read these figures compared
(ΔQ=QGIC-QSS) with difference in the order of 10-20% to Fig. 4 and 6, but we can observe the same behavior with a
underestimation for this transformer. On Fig. 10 the reactive DC flux offset in leg, yoke and outer-limb. The magnitude of
power Q is shown to be directly proportional to GIC. outer-limb flux is reduced once saturated, and the top of the
Q is calculated here using the traditional definition of positive peak is limited due to the unidirectional saturation.
Air-path inductance effect is noticeable only once the outer- differential current is a direct effect of the core saturation and
limb becomes fully saturated and contains both a DC and AC increased distorted magnetization current and flows in the
components. Air-path inductance has only a minor effect on neutral of the transformer.
the magnitude of the leg fluxes. The harmonic analysis of the differential current in Fig. 15
500 shows the presence of all even and odd harmonics typical of
[Wbt]
unidirectional waveforms and similar to those typically found
375
Leg in transformer inrush currents.
250
70
125 Outer-limb without Lair
with Lair
60
0 DC
1st
-125 Yoke 50 2nd
3rd
-250 4th
40 5th

[A]
-375
6th
7th
30
-500
0 15 30 45 60 75 [s] 90
20
Fig. 11: Flux-linkages in different core sections with AC source and
neutral voltage applied (100 A GIC) according to Fig. 1 without any
air-path inductances. 10

500 0
0 10 20 30 40 50 60 70 80 90
[Wbt] [s]
375
Leg Fig. 14: Evolution of harmonic content of differential phase current,
250 without and with (500mH) air-path inductances for 100 A GIC.

125 Outer-limb
80

0
Air-path 60
without Lair
with Lair
Yoke
[A]

40
-125

20
-250
0
DC 1 2 3 4 5 6 7
-375
Harmonic at Saturation Equilibrium

-500
Fig. 15: Harmonic content of differential phase current at saturation
0 15 30 45 60 75 [s] 90 equilibrium, without and with (500mH) air-path inductances for 100 A
Fig. 12: Flux-linkages in different core sections with AC source and GIC.
neutral voltage applied (100 A GIC) according to Fig. 1 with air-path
inductances added (500 mH). V. DISCUSSION

300
In addition to the increased reactive power demands, power
utilities are concerned with the protection relay false tripping
[A]
with air-path
due to GIC. Relevant protection schemes that may be affected
250
inductance by GIC are differential protection and high-impedance
200 ground-fault protection installed on HV side.
How the GIC will influence the protection of the trans-
150 former is a complex issue. To get a clear answer saturation of
without air-path
the current transformers should be taken into account in the
100 inductance
simulations and the simulated waveforms should be applied to
a real protection relay.
50
Generally, the differential protection of the transformer is
set for clearing internal short circuits or external short circuits
0
0 15 30 45 60 75 [s] 90 close to the transformer. Differential protections have
Fig. 13: Transformer differential phase current, with (500mH) versus harmonic stabilization normally on the 2nd and 5th harmonics.
without air-path representation for 100 A GIC.
2nd harmonic stabilization may be only active at transformer
energization. For higher values of these harmonics the trip is
Fig. 13 shows that the influence of the air-path inductances
blocked. A typical relay characteristic with harmonic traces as
on the differential current is significant. For the same GIC
function of the GIC is shown in Fig. 16. Different level of GIC
current (same DC component) Fig. 14 and Fig. 15 show a
gives equivalent harmonic curves with the tip function of the
considerable difference in the harmonic content of the
GIC level. In Fig. 16, only unusual high level of GIC may
transformer differential current. In case of GIC, the
trigger the differential protection. However, protections may for providing access to transformer data.
still trigger as a result of saturation in the current transformers.
The system-wide protection response to GIC should be further VIII. REFERENCES
investigated. [1] P. Price, “Geomagnetically induced current effects on transformers,”
Power Delivery, IEEE Transactions on, vol. 17, no. 4, pp. 1002-1008,
Trip Area Trip Area Blocking Area
2002.
2nd harm 5th harm
500 [2] Walling, R. A.; Khan, A. N. “Characteristics of transformer exciting-
500
0.5
current during geomagnetic disturbances”, IEEE Transactions on Power
500 500
Delivery, Vol. 6, Issue: 4, pp. 1707-1714, 1991.
Ph.A/C
[3] R. Pirjola, “Geomagnetically induced currents during magnetic storms,”
Ph.B
0.4 IEEE Trans. Plasma Sci., vol. 28, no. 6, pp. 1867–1873, Dec. 2000.
Ph.B 2nd harm. [4] Kappenman J.G.: ‘Transformer dc excitation field test and results’. IEEE
n

Ph.A/C PES Special Panel Session Report, 90 TH0291-5 PWR, 12 July 1989
/I

0.3
5th harm.
1-diff

[5] Kappenman J.G., ALBERTSON V.D., MOHAN N.: ‘Current

transformer and relay performance in the presence of geomagnetically-
I

200 200 induced currents’, IEEE Trans. Power Appar. Syst., 1981, 3, (PAS-100),
0.2 200 200
pp. 1078–1088
[6] B. Zhang, L. Liu, Y. Liu, M. McVey and R. Gardner, “Effect of
100 100 100 100
0.1 geomagnetically induced current on the loss of transformer tank,”
50 50 50 Electric Power Applications, IET, vol. 4, no. 5, pp. 373-379, 2010
20 [7] Seyed Ali Mousavi, “ lectromagnetic Modelling of Power Transformers
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 with DC Magnetization” , Licentiate Thesis in Electromagnetic
Ih-diff / I1-diff Engineering, TRITA-EE 2012:057, ISBN 978-91-7501-537-8, Royal
Fig. 16: Differential current protection: 2nd and 5th harmonics blocking Institute of Technology (KTH), , Stockholm, Sweden 2012
characteristics. Harmonic traces as function of the GIC numbered in [8] Xuzhu Dong; Yilu Liu; Kappenman, J.G., “Comparative Analysis of
Ampere along the curves for phases A/C and B. Exciting Current Harmonics and Reactive Power Consumption from
GIC Saturated Transformers”, Power Engineering Society Winter
Meeting, pp. 318 – 322, vol.1, 2001
The modeling approach and air-path inductance estimation [9] Jinxia Yao; Min Liu; Changyun Li; Qingmin Li, “Harmonics and
method presented in this paper can be applied to three-limb Reactive Power of Power Transformers with DC Bias” Power and
transformers. Three-limb transformers do not have an iron Energy Engineering Conference (APPEEC), 2010 Asia-Pacific, Pp. 1–4,
2010
zero-sequence path, therefore only the air-paths determine the [10] Hongzhi Li; Xiang Cui; Tiebing Lu; Zhiguang Cheng; Dongsheng Liu,
response of GIC and should be estimated with great accuracy. “An improved magnetic circuit model of powertransformers under DC b
Nevertheless, three-limb transformers have the advantage that ias excitation”, Electromagnetic Compatibility (APEMC), 2010 Asia-
air-path inductances can be measured from open-circuit zero- Pacific Symposium on, pp. 806 – 809, 2010
[11] Masoum, Mohammad A.S.; Moses, Paul S., “Influence of
sequence tests. Geomagnetically Induced Currents on three-phase power transformers”,
Power Engineering Conference, 2008. AUPEC '08. Australasian
VI. CONCLUSIONS Universities, pp. 1–5, 2008
[12] E. F. Fuchs, Y. You, and D. J. Roesler, "Modeling and simulation, and
The GIC response is shown to be significantly affected by the their validation of three-phase transformers with three legs under DC
air-path inductances even in five-limb cores. Transformer bias," IEEE Transactions on Power Delivery, vol. 14, pp. 443-449,
models used for the investigation of GIC phenomenon should 1999.
[13] E. F. Fuchs and Y. Yiming, "Measurement of Lambda-I characteristics
have at least the following characteristics: of asymmetric three-phase transformers and their applications," IEEE
- topology correct core representation; Transactions on Power Delivery, vol. 17, pp. 983-990, 2002.
- accurate representation of highly saturated state by taking [14] D. Tousignant, L. Bolduc, A. Dutil, “A method for the indication of
power transformer saturation”, Electric Power Systems Research,
into account the air-core inductance; Volume 37, Issue 2, May 1996, Pages 115–120
- modeling of the flux paths outside by air-path inductances. [15] Picher, P.; Bolduc, L.; Olivier, G., “Acceptable direct current in three-
The most critical core sections are the outer-limbs as they phase power transformers: comparative analysis”, Electrical and
Computer Engineering, 1997. Engineering Innovation: Voyage of
become highly saturated in the event of GIC. The ratio Discovery. IEEE 1997 Canadian Conference on, pp. 157-160
between the fully saturated outer-limb inductance and the air- vol.1, 1997
path inductances determines the extent of flux forced outside [16] H. K. Høidalen, B. A. Mork, F. Gonzalez, D. Ishchenko, N. Chiesa:
the core. FEM seems to be the only viable alternative for the “Implementation and verification of the Hybrid Transformer model in
ATPDraw”, Electric Power Systems Research 79 (2009), 454-459.
calculation of air-path inductances in five-limb transformers. [17] Mork, B.A.; Ishchenko, D.; Gonzalez, F.; Cho, S.D.: "Parameter
When relevant, transformer manufacturers should be able to Estimation Methods for Five-Limb Magnetic Core Model", IEEE
provide the value of air-path inductance, and empirical Transactions on Power Delivery, Vol. 23 , pp. 2025-2032, 2008.
methods for the estimation of this parameter should be
developed as FEM modeling is inconvenient in most
circumstances.
The accurate transformer model presented in this paper is
an important basis for further work on protection response.

VII. AKNOWLEDGMENTS
The authors wish to thank ABB Ludvika and Hafslund Nett

View publication stats