Comparative Analysis of Exciting Current Harmonics and Reactive Power

Consumption from GIC Saturated Transformers

Xuzhu Dong, Yilu Liu, John G. Kappenman

Department of Electrical Engineering Metatech

WI& Su
Blacksburg, Virgini~ 24061-0111 Duluth, MN, USA
USA

Abstract: Geomagnetically induced currents (GIC) can cause severe MVar consumption directly with the given GIC, which is
transformer half-cycle saturation. This paper introduces a simplified helpful for on-line GIC monitoring. The effect of the
method based on the equivalent magnetizing curve of the transformer transformer parameters on the GIC induced saturation is also
to estimate harmonic currents and MVar with only the given the GIC discussed, such as the voltage, MVA, and the normal exciting
and nameplate and core design information of the transformer. The
current.
simulation method is validated with the field results. The simulation
methods are validated with observational and field test data for
transformers experiencing half-cycle saturation. II. DESCRIPTION OF THE SIMPLIFIED APPROACH

Keywords: Transformer, GIC, Saturation, Harmonics, Reaetive A. Assumptions

power
The following assumptions are given to simplifi the
I. INTRODUCTION problem.
1. The nameplate information of the transformer is known.
Geomagnetically induced current (GIC) is a quasi-direct including the rated voltage and MVA, auto or normal, phase
current, and can result in a high level of half-cycle saturation numbers, core configuration, and the typical excitation current
when present in a transformer. A transformer with the half- percentage.
cycle saturation becomes a rich source of even and odd 2. The input GIC is known and divides equally between
harmonics and draws significant inductive MVar from the three phases, either in single-phase transformer banks or
power system ‘1]. Geomagnetic disturbances can produce three-phase transformers. In the case of autotransformers, it is
wide-spread flows of geomagnetically induced currents especially important that the GIC flowing in all terminals
simultaneously throughout a power network. Therefore in (including ground or neutral) be known so that an effective
order to understand the impacts of geomagnetic disturbances GIC in the transformer can be derived.
on power grids, it is important to accurately model the 3. The AC and DC flux only flows in the core and the
behavior of multiple individual transformers experiencing leakage flux is ignored, except for the transformer with three-
differing levels of GIC in a large scale model. In this case, phase, 3-legged, core form. Iron loss and copper loss are also
the simplified models described by the authors are limited to ignored, and then the formula below can be obtained.
modeling the AC reaction of the transformer due to saturation Assume that the normal phase voltage u(t) and induced
on the network. These saturation impacts being increased electro-motive force (EMF) e(r) of the exciting winding are
power demands and increased harmonic current production. fundamental-frequency positive-sequence and known, the AC
Internal heating, another potentially important problems, is linkage yin(t)resulting in e(t) is
not considered in thk model approach.
Staged tests have been performed to study the harmonics
yac (/)= –j e(t)dt = -J(-u(t))dt = @‘1 JZJl sin(r3t) (1)
and reactive power consumption during transformer saturation
’278]. Several simulation methods have also been developed to
calculate harmonic currents and MVar intake of transformers UI is the RMS value of u(t). If the DC linkage VdCinduced by
resulting from GIC. These include the FEM based models ‘3], GIC exists, then the total linkage tit)in the transformer core
and the magnetic circuit models ‘4].These methods are usually can be represented by the exciting current i(t ).
accurate but complicated and require detailed design
information that is generally not available, tit)
= y=(t)+ = F(iw(t)+Idc)
~c = F(i(t)) (2)
To estimate harmonic currents and MVar with only the
given GIC and the nameplate information of the transformer, Here ire(t) is the AC component of i(t), Id. is the value of GIC
it is necessary to look for simplified methodsthat will provide per phase, the fimction F is the equivalent vi magnetizing
sufilcient accuracy for large-scale network models of storm curve of the transformer.
impacts. Results of the model are presented in this paper is to Provided that V=(t), Z&, and the function Fare known, W,.
validate a simplified approach for transformers with different and iw(t) could be calculated from Eq.2 by using an iteration
core design. Harmonics and the MVar consumption of algorithm called binary search method ‘5]. This method is
transformers with different core design are analytically based on the fact that DC component in the exciting current
compared, and a simple formula is obtained to evaluate the

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 318
Authorized licensed use limited to: University of Otago. Downloaded on September 24,2023 at 22:42:26 UTC from IEEE Xplore. Restrictions apply.
 i(t) is equal to the applied GIC, I&. FFT is used to find 2. Three-phase, 3-legged, core form. Similar to the case of
harmonic currents from i=(t), and the MVar Q is given below three-phase, shell form, the different values of the slope kz, k3,
the second knee CZ, and the scaling factor of GIC are
Q = 3* Ul”Il (3) determined by test results.
3. Three-phase, 5-iegged, core form. Supposed that the
11is the fundamental component of i(t). positive-sequence AC flux only exists in the main legs and the
magnetic path per phase can be roughly shown in Fig.2 (a).
B. Treatment of auto transformers The DC flux flows between the side leg and the main leg as
shown in Fig.2 (b). Here S is the cross-section area of the
Auto transformers can be modeled as two-winding main leg, and that of the side leg is half of the main leg. L is
transformers as long as an effective GIC or DC current&c is the leg height. The equivalent AC and DC magnetizing curves
used for the unequal flow of GIC in the series and common are shown in Fig. 1 (a). The AC curve can be found according
windings ‘6]. to the no-load V-I curve of the transformer, and the DC curve
is obtained below.
Z* =fI&,+z#(k+l) (4) Provided that FI is the AC magnetic-motive force (MMF)
in Fig.2 (a), the first slope kw of the AC magnetizing curve is
Where Id., and I& are the GIC flowing in the series winding given below.
and the common winding, respectively. k = VH/VL-1. VH and
VL are the terminal voltages on the HV and the LV side of the F,= NIW = HIL + H2L
auto transformer. In the analysis below, only two-winding =/d*@S+,ul*y/(2S) =,Ld/S*(l.5~ (5)
transformers are discussed assuming that auto transformers
have already been represented by two-winding transformers. k== ~(NI=) = (2/3) *S/(/d) (6)

C. Simpll~ed algorithm for single-phase transformers Where Ix is the AC exciting ctu?ent, HI and Hz are the
magnetic strength of the legs, v is the AC linkage, N is the
The process of estimating harmonic currents for the single- turn number of the exciting winding. The effect of yokes is
phase transformer is straight-forward since it involves only ignored. F2 is the DC MMF in Fig.2 (b), and ~& is the DC
one common main flux path that can be represented by a non- current per phase, then the first slope of the DC magnetizing
linear inductance. The magnetizing curve VZ can be given curve kdcis shown below.
directly from the no-load voltage-current V-Z curve of the
exciting winding according to Eq. 1, either core form or shell
form. The typical magnetizing curve can be replaced by a
proxy using a piecewise linear representation as shown in
Fig. 1 (a), k, and kz are the slopes of the lines. As a rule of
thumb, ].1 is equal to 1.1* fi& at the knee of the
magnetizing curve. l=. is the RMS value of the normal
exciting current.

D. Treatment of three-phase transformers

Since the AC and DC flux path are different for three-phase

transformers, the different equivalent magnetizing curve
should be given for different core designs. Here the DC flux (a) Single phase (b) Three-phase, shell form
per phase is assumed to be equal for the sake of
simplification. Since very few three-phase transformers with Fig. 1 The equivalent magnetizing curve of the !xansformes
7-legged, shell form are in use, the analysis about this kind of

m
transformers is not considered below.
1. Three-phase, shell form. The equivalent @ curve is
represented by five straight lines, other than three lines, in
s S12 L
order to roughly describe the complicated AC and DC flux
H,
coupling in the core. As shown in Fig. 1 (b), the slope kl and
+&,:
the tirst knee Cl are determined according to the rated voltage
and the normal exciting current similar to the case of the (a) AC magnetic path (b) DC magneticpath
single-phase transformer. The slope kz, k3, and the second
knee Cz are determined by field results. Since the path of DC Fig.2 The simplified magnetic path of the three-phase, 5-
flux induced by GIC is different from the AC flux path, a Iegged, core form transformer
scaled GIC which would have the same effect as the original
GIC in the core is considered in Eq.2.

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 319
Authorized licensed use limited to: University of Otago. Downloaded on September 24,2023 at 22:42:26 UTC from IEEE Xplore. Restrictions apply.
0-7803-6674-3/00/$10.00 (C) 2000 IEEE
 F,= NI&= M* @Y + /d* y/i(O.5*S) = ,WS*(3 y/) (7) Table 1 The comparison of the second harmonic current of transformers with
different core design between the simulated and measured results
k&= ~(NI~J = (1/3) *S/@Z)= kJ2 (8)

Refer to Fig.1 (a), the linkage Y at the knee of the DC \-) ..

Measured
(b) 2.(
magnetizing curve in Fig.2 (b) is the same as the the case in results
@MS, A) () 1.41 I
Fig.2 (a). The second slopes of the magnetizing curves in (:)
Fig.2 (a) and (b) are determined according to test results. I
() 11 ‘
Simulated
The DC flux @& is first found by using Eq.2 and the results
(:) 2.
magnetizing curve in Fig. 1 (a). The DC current in Eq.2 should () 1.
@MS, A)
(:)
be one and a half of GIC per phase because the half of DC
() 4.
flux in Phase B contributes to the magnetic circuit in Fig.2
Deviation (;) 7.
(b). Assume that y=(t)in Eq.2 is still the normal operating (*loo%) (c) 13
AC flux. The exciting current i(t) is given by applying the (d) “
magnetizing curve in Fig.2 (a) with the obtained @& and
(a) The measured transformer is 360 MV~ 500/230 kV, single-phase, shell
normal ~W(t). form ‘s].The exciting current is 0.001 p.u.
The normal exciting current leXCcan be estimated from the (b) The measured transformer is 200 h4V~ 230/115 kV, three-phase, shell
rated load current of the transformer, usually fall into 0.2% - form “]. The exciting current is 0.0026 p.u.
0.5’% for large power transformers. (c) The measured transformer is 200 MV~ 230/115 kV, three-phase, 3-
[sl me exciting Cwent is 0.0026 P.”.
Iegge& core form
(d) The measured tramfonner is 750 MV~ 525/303 kV, three-phase, 5-
III. VERIFICATION OF SIMULATION RESULTS legged, core form .
[9]me exciting cwent is 0.0004 p.w

The simulation results were compared with test results to

veri~ the feasibility of the simplified method. Because of the ,93, 18,

presence of load, the exciting current of the transformer

becomes superimposed upon the load current, and the
fundamental component of the exciting current is difficult to
measure accurately ‘8]. As shown in Table 1, only the second
harmonic current is compared with the test results for
different core designs, except for the single-phase core and
shell form ‘8’9]. The deviation between the simulated and test
results is also given for different GIC values. The maximal
deviation is within 30% and most of them are within 10%. Time (ins) Frequency (W)
The deviation for single-phase shell form and three-phase 5- (a) me exciting c~ent wavefo~ (b) The exciting current harmonics
legged core form is smaller than for three-phase shell form
Fig.3 The simtdated exciting current for a GIC of 11.5 A per phase
and three-phase 3-legged core form. (single-phase, core form transformer)
Fig.3 shows the simulated exciting current waveform and
the tlequency spectrum with a GIC of 11.5 A per phase for
the single-phase core form transformer. The measured Our study shows that the effect of GIC is similar for the
transformer is 400 MVA, 5121242 lcV. The exciting current is single-phase shell form and single-phase core fore, if the
0.32 A ‘6].The simulated result agreed also well with the test above parameters are used, so the core designs compared
result. It shows that the simplified method is feasible with include: single-phase, three-phase shell form, three-phase 3-
adequate accuracy for all core designs. Iegged core form, three-phase 5-legged core form. The GIC
range of interest per phase is from O to 40 A with the
IV. CASE ANALYSIS AND COMPARISON increment of 1 A. Various outputs for each core design and
different GIC are obtained and compared, such as the exciting
A series of simulations were performed to compare the GIC current harmonics, MVar consumption.
effect on transformers with different core design based on the
simplified method. For easy comparison in the following case A. The exciting current harmonics
analysis, the transformer parameters are set constant for
different core designs. The variation of the exciting current peak and harmonics
with the increasing GIC are shown in Fig.4 and Fig.5. The
MVA: 300 MVA exciting current peak increases nonlinearly with GIC for all
Voltage: 500/230 kV core designs, and the increasing rate is also different. For a
Frequency: 60 Hz fixed GIC, the exciting current peak for single-phase is
The exciting current 0.42 A always the highest, and a decreasing order for other core
designs is three-phase 5-legged core form, three-phase 3-
legged core form, three-phase shell form.

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 320
Authorized licensed use limited to: University of Otago. Downloaded on September 24,2023 at 22:42:26 UTC from IEEE Xplore. Restrictions apply.
0-7803-6674-3/00/$10.00 (C) 2000 IEEE
 t~2, and t2= T14 + tJ2. GIC per phase can be obtained by the
3W
triangle area S divided by the period of T, that is,
~.

$2?0
— Single-phase GIC = S/T= (A@/2)/T (9)
b ------ Three-phase, shell form
Sm ~ /
The exciting current i(t)and its Fourier series are given
~ Three-phase, 3-legged
below.
‘El // – ~=-ph~e$,’==d
‘4(l+(t-~~).~d) t,<t<T~
i(t) = .4(1- (t-T~).~d) %<t+
(10)
GIC per phase (A) [
0’
Fig.4 The variation of the exciting current vs the input GIC per
phase
I others

In =+ f~i(t). exp(-jrcoot) d
2

(11)

[1
sin(nq f)
=GIC. exp(-jnroo $).
td
rrc&2
~
GIC per phase (A) GIC per phase (A)
(a) Single-phase (b) Three-phase, shell form
Where @ = 27dT, n = 1,2,3 Eq. 11 means that not only the
,6 35 fundamental component 11, but also all other harmonics are of
2,, good linear relationship with GIC. The magnitude of the

E
?. /

g’: harmonics falls off as l/n. In fact, since the duration td does
also vary with the increase of GIC, this results in the
.: e nonlinear relationship between the higher order harmonics
04 and GIC. Thk has been verified by field measurements ‘2].
E
z
~2 Fig.5 also shows that the even harmonics increase
0 >0203040! dramatically with the increase of GIC, different from the case
GIC per phase (A) GIC per phase (A) of the transformer over-excitation. Half-cycle saturation,
caused by the presence of GIC, will produce significant levels
(c) Three-phase, 3-Iegged, (d) Three-phase, 5-le~ed,
core form core form of even harmonics. The detection of even harmonics serves
as a reliable indicator of the presence of GIC and half-cycle
Fundarnerrtrd-frequency
saturation.
~ 4ti harmonics
------ 2“dharmonics +++++ 5ti harmonics
B. Reactive power consumption
------- 3rdharmonics eegeeo 6ti harmonics
The MVar consumption of the transformer is calculated by
Fig.5 The relationship of the exciting current hwrnonics and GIC for
using Eq.3. As shown in Fig. 7, the variation of the MVar
transformers with different core design
consumption is also linear with the increase of GIC for the
sake of the fimdamental component of the exciting current.
The harmonics variation in Fig.5 has the same trend as the The MVar drawn by the transformer can be easily estimated
case of the exciting current peak. It should be noted that the by using the following formula according to Fig.7.
fundamental and second harmonic currents are linear with
GIC in the calculated range for all types of core design. The Q (MVar) = kl*GIC + Q, (12)
harmonics become nonlinear with the increase of GIC if the
harmonics order is higher. This can be explained below. Here QO is the reactive power consumption from the normal
As shown in Fig.3 (a), if GIC is five times more than the exciting current and the value of GIC is in neutral. The value
peak of the normal exciting current, the normal exciting of kl is given in Table 2 for different core designs.
current can be ignored and the exciting current waveform cart
be approximately represented by a periodical triangle wave as V. CONCLUSION
shown in Fig.6. Here A is the exciting current peak, f = UT is
the fundamental frequency, k] and kz are the slopes. The To estimate harmonic currents and MVar with only the
duration of the trirtngle puke td iS equal to tz-tl, then tl= T/4 - given GIC and the nameplate information of the transformer,

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 321
Authorized licensed use limited to: University of Otago. Downloaded on September 24,2023 at 22:42:26 UTC from IEEE Xplore. Restrictions apply.
0-7803-6674-3/00/$10.00 (C) 2000 IEEE
 The authors gratefully acknowledge the contributions of
—.
L _____________________
National Science Foundation (NSF) and Metatech financial
support of this work.
z
.-
.=
? k, VII. REFERENCES
H
w [1] Vernon D. Albertsou Geomagnetic disturbance causea and power
system effects, IEEE Special Panel Session Repo@ IEEE PES Summer
Meeting Jrdy 12,1989, Long Beac~ Califomi~ USA pp.3-9
t, T/4 ~2 T12 b
T [2] RL. Lesher, C.L. Wagner, W.E. Feero, SUNBURST GIC Network
Time EPRIRepo~RP-3211-01, Nov. 1993
[3] Shu LL Yilu LiL A fundamental analysis of @rnsforrner GIC
Fig.6 The approximate wavefotm of the exciting current in case of
magnetization using the finite elements approach, Proceedings of the
the transformer saturation
American Power Conference, A@ 1991, Illinois Inst. Technol, Illinois,
USA pP.1 173-1178
[4] RA Walling AH.Khaq Characteristics of transformer exciting-current
during geomagnetic disturbances, IEEE Trans. on Power Delivery, vol.
6, no. 4, 1991, pp. 1707-1714
[5] W. XW T.G. Martinic~ J.H. %wad~ Y. Mansour, Harmonics from SVC
transformer saturation with direct current offset IEEE PES Summer
3.
Meetin& 404-4 PWRD, 1993, pp. 1-7

&
3 ------ Three-phase, shell form
~= [6] D.H. Boteler, RM. Shier, T. Watanabe, RE. Hori@ Effcts of
al” +++++ Three-phase, 3-legged geomagnetically induced currents in the B.C. hydro 500 kV system,
.: ,6
0 ~ Three-phase, 5-legged IEEE Trans. on Power Delivery, vol. 4, no. 1, 1989, pp.818-823
~ ,,
[7] Ymg Kening, power engineering Press of Xi an Jiaotong Univ., 1984,
&,
pp.173
o,,o,, mss am [8] John G. Kappc- Transformer JX excitation field test& results,
GIC per phase (A) IEEE Special Panel Session Repoz IEEE PES Summer Meeting, July
12, 1989, Long Beach, California, USA pp. 14-22
Fig.7 The variation of the MVm consumption vs the input GIC [9] GeneraJ Electic Company, High-voltage direct-current converter
per phase transformer rnagnetics, EPRI EI..-434O, DC. 1985

VIII. BIOGRAPHIES
Table 2 The value of k, and kz for different core designs
Xuzhu Dong is pursuing his second Ph.D. degree in Power Systems at
Vbginia Tech. He received his BS, MS, and PHD degrees from Tsinghua
University, China, in 1993, 1998. His current research interests are power
system transient power equipment monitoring and diagnosis, Internet
applications in pewer system. H]s E-mad address is [email protected]

Yihr Lhr is an Associate Professor of Electrical Engineering at V@inia

Tech. Her current research interests are Internet/IT applications in power,
power quality, power system equipment modeling and diagnoses, and
a simplified method based on the equivrdent magnetizing FACTS/power electronics. Dr. Lkr is the recipient of the 1993 National
curve of the transformer is developed for transformers with Science Foundation Young Investigator Award and the 1994 presidential
Faculty Fellow Award. Her E-mail address is [email protected]
different core design. Compared with test results, the
simulation method is feasible with adequate accuracy. John G. Kappenman is a 1976 graduate in Electrical Engineering from
The effect of GIC on transformers with different core South Dakota State University. He joined Mmesota Power (1977-1998)
design is analyzed based on the simplified method. The holding a number of senior management and tectilcal positions. br 1998 he
joined Metatech m the Manager of their Applied Power Solutions Division.
relationship between the harmonics, the MVar consumption His current research interests are Lightning and Space Weather impacts on
and GIC are illustrated. Simulation shows that the electric utilities. He is a IEEE Senior Member and the Past Chairman of the
fimdamental component of the exciting current and the Transmission and Dkribution Cornmittee(1994-1996). He has published
reactive power are linear with GIC, making the evaluation of over 30 papers in a variety of subject areas. He is a recipient of the IEEE
Waker Fee Outstanding Young Engineer Award, the IEEE Prize Paper
the MVar consumption directly with the given GIC possible. Award, the Westinghouse Nikola Tesla Awad and two EPRI Innovator
Awords. Hk end address is JKwmenm@aol corn
VI. ACKNOWLEDGMENTS

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 322
Authorized licensed use limited to: University of Otago. Downloaded on September 24,2023 at 22:42:26 UTC from IEEE Xplore. Restrictions apply.
0-7803-6674-3/00/$10.00 (C) 2000 IEEE