For as long as there have been conductive networks on our planet's surface, there have been geomagnetically
induced currents (GICs) (Boteler & Pirjola,1998). These currents, which are caused by variations in the Earth's
magnetic field, flow through grounded conductive systems such as power grids, moving between the conductive
power lines and the earth via transformers (Price,2002). Due to the damage and disruption GICs can cause within
power grid infrastructure (Molinski,2002) such as that seen in Quebec during the March 1989 geomagnetic storm
(Bolduc,2002), modeling and measuring of potential currents is seeing increased interest due to grid operation
safety concerns (Kelbert,2020; Oughton etal.,2017).
Research into GICs in Austria began in 2014 in a study initiated by the Austrian Power Grid AG (APG) and
summarized in Halbedl etal.(2014). The aim of this first attempt was to investigate possible DC currents in the
Austrian high voltage power transmission network. For APG, measurements of DC are important to study the
nature of the currents and possible effects on transformers, as well as the impact of DC on different equipment
such as instrument transformers and protection devices. The data are also used as a planning basis for the design
criteria of new transformers.
The section of the Austrian power grid under investigation in the 220 and 380kV levels has 56 interconnected
substations with a total length of 6,965km of power lines and 87 transformers, spanning an area of 84,000km2.
After the initial test measurements in 2014, more DC measurement devices were installed with the aim of a
long-term measurement campaign, and the first data analysis was carried out in Halbedl etal.(2016) at the
Institute of Electrical Power Systems at Graz University of Technology (IEAN, TU Graz) and later in Bailey
etal.(2017,2018) at the Conrad Observatory for geomagnetic field measurements (ZAMG). The studies at each
Abstract Geomagnetically induced currents (GICs), a result of solar wind interaction with the Earth's
magnetic field and the resistive ground, are known to flow in power transmission grids, where they can
lead to transformer damage and grid operation problems. In this study we present an analysis of five years
of continuous GIC measurements in transformer neutral points in Austria. Seven self-designed stand-alone
measurement systems are currently installed in the Austrian 220 and 380kV transmission levels, measuring
currents up to 25A. We identify recurrent geomagnetic activity in the measurements, and also find man-
made sources of low frequency currents using frequency analysis. In order to support the transmission grid
operators, two GIC simulation approaches are used to simulate GICs in the power grid. The first model uses
measurements to derive the sensitivity of the location to northward and eastward geoelectric field components
(which requires no detailed grid data), and the second model uses the detailed grid model to compute GICs
from a geoelectric field. We evaluate two geomagnetic storms from September 2017 and May 2021 to discuss
the effects of GICs on the power transmission grid and its assets.
Plain Language Summary During geomagnetic storms, rapid changes in the Earth's magnetic field
induce an electric field in the ground, may drive currents in the power grid. These are called geomagnetically
induced currents (GICs) and they can lead to power grid operation problems and even transformer damage. In
this study we present learning's from five years of GIC measurements in Austria, which have been carried out
in seven different transformers in the grid. Some power grid transformers show larger susceptibility than other
transformers to magnetic field variations accompanied by larger GICs. We also identify some of the sources of
noise in the data such as a city subway system, and investigate two geomagnetic storms from September 2017
and May 2021 in more detail.
ALBERT ET AL.
© 2021. The Authors.
This is an open access article under
the terms of the Creative Commons
Attribution License, which permits use,
distribution and reproduction in any
medium, provided the original work is
Analysis of Long-Term GIC Measurements in Transformers in
D. Albert1 , P. Schachinger1 , R. L. Bailey2 , H. Renner1 , and G. Achleitner3
1Institute of Electrical Power Systems, Graz University of Technology, Graz, Austria, 2Conrad Observatory, Zentralanstalt für
Meteorologie und Geodynamik, Vienna, Austria, 3Austrian Power Grid AG, Vienna, Austria
• Measurements of geomagnetically
induced currents (GICs) in power grid
substation transformers have been
carried out since September 2016 in
• We summarize the measurements
until now and discuss data quality and
sources of noise
• A statistical analysis of GIC
measurements, a comparison of
simulation models for two storms
and an attempt to a risk assessment is
Albert, D., Schachinger, P., Bailey, R.
L., Renner, H., & Achleitner, G. (2022).
Analysis of long-term GIC measurements
in transformers in Austria. Space
Weather, 20, e2021SW002912. https://
Received 14 SEP 2021
Accepted 22 DEC 2021
Conceptualization: D. Albert
Formal analysis: D. Albert
Investigation: D. Albert
Software: D. Albert
Space Weather Impacts on
Electrically Grounded Systems
at Earth's Surface
1 of 16
ALBERT ET AL.
2 of 16
institute had different focus areas: the TU Graz looked in detail at the different sources of DC in the power grid
and the effects on transformers, while the ZAMG focused on geomagnetic variations and the scales of possible
GICs both past and future. This work continues in Albert etal.(2019,2020).
There are now five years of measurements of DC in transformers at various locations in Austria. Worldwide, there
are still few countries with GIC measurements spanning long time periods and multiple locations, with some ex-
amples being New Zealand (Rodger etal.,2017), China (Zhang etal.,2015), more recently the USA (Kellerman
etal.,2021) and Finland (Pirjola,1989).
In addition to the measurements, two GIC simulation models for the Austrian power grid have been developed
based on two different geoelectric field modeling approaches.
The first model, from the IEAN, is based on the plane-wave method in combination with the power grid model.
The simulation can be interfaced with a graphical user interface (GUI) and is publically available with sample
data at https://github.com/P-Schachinger/LFC_simulator. The second method is also based on the plane-wave
method, but operates without the detailed power grid data. Instead of the power grid data, factors for the sensi-
tivity of each substation are derived from past transformer neutral current measurements. A third model from the
ZAMG developed in the past, which is not presented here, is based on calculation of the geoelectric field on a
surface with a surface conductive thin-sheet, and can be found at https://github.com/bairaelyn/GEOMAGICA. A
comparison of different models with different degrees of detail was performed in Bailey etal.(2018) and sum-
marized in Table 2 in the same publication. The results reveal only slight improvements with increasing degree
of detail of the Earth model with the thin-sheet approach. Therefore, a higher degree of detail do not justify the
effort for the ground modeling. However, due to the presence of the Alps, these effects could be larger in the
western part of Austria, where there have been fewer measurements to date. The absence of any highly conductive
coastlines makes Austria a far simpler case than for example, the UK or Sweden.
In this paper the results of an analysis of the entire data set of DC measurements and summarize the lessons
learned in the five years since the measurements started.
2.1. DC Measurement System
The IEAN transformer neutral point current (NPC) measurement system (version 3) is a self-engineered stand-
alone and remote-controllable data acquisition system. The measurement system uses an active low pass filter
with a cut-off frequency of 0.7Hz to damp for example the 50Hz power system frequency, meaning the sampling
rate of the data acquisition can be reduced to 1Hz. An active second order low pass filter in Sallen-Key design
is preferred over a passive filter in order to reduce the filter components. This low sampling frequency allows to
use a low-cost single-board computer, such as a Raspberry Pi.
The measurements of currents are done with a Hall effect closed-loop zero flux current transducer. A shunt
resistor in series to the transformer neutral would change the impedance and therefore the GIC amplitude. The
shunt would also be large in size because it would need to carry a high short-circuit current in the case of a line-
to-ground fault. The measurement system has a guaranteed accuracy of 2%±1mA for DC in the range of ±0.1A
to ±25A. The actual measurement accuracy for DC in the range of 1–25A is below 0.2%±1mA. A technical
description of the second version of the measurement system can be found in Halbedl(2019).
The measurement data is sent to an external server at the TU Graz. Before use, the data is cleaned and missing
data or absolute values above 25A are automatically flagged. The missing values are interpolated with the
"Piecewise Cubic Hermite Interpolation" (PCHIP) method (Fritsch & Carlson,1980), which results in lower
amounts of overshooting in comparison to other interpolation methods. Considering the effect of interpolation in
the frequency domain, the PCHIP method provides the least change in the frequency spectrum of the measured
2.2. GIC Measurements
At the time of writing (September 2021), seven transformer neutral points in the 220 and 380kV transmission
levels across Austria are equipped with our GIC measurement system, as depicted in Figure1. The measurement
ALBERT ET AL.
3 of 16
locations are named according to order of setup going from #01 to #08 (where #06 is not in use). Measurements
started in September 2016 with one measurement system near Vienna in a 380kV transformer neutral. Figure2
gives an overview of the runtimes of all measurement systems. Client #02 was first situated in a 380kV neutral
point in eastern Austria, before it was moved to a 220kV transformer neutral point in central Austria. All follow-
ing descriptions of #02 refer to data measured at its second location.
In much of the data up until 2021 there was a cut-off point in the measurements in the negative direction of 3.5A
(but not in the positive direction). The saturated measurement periods/days are excluded in the data statistics and
further analysis. This is due the electronic design of the former transformer neutral point current measurement
system. The systems were installed in such a way that they were able to meas-
ure up to 25A in one direction and up to 3.5A in the other direction of the
transformer neutral point current. With a new electronic design, established
at all measurement locations from mid-2020 to mid-2021, the systems are
now able to measure positive and negative currents up to an amplitude of
The GIC events are detected with an automated algorithm. This is required
because short-duration peaks, caused by switching events and faults in the
power grid are also recorded. A GIC event is defined as a current with a
peak prominence of at least the mean value plus the standard deviation of the
analysis time span. The half width duration of the prominence needs to be at
least 100s. An automated algorithm for the detection of geomagnetic storms
based on the analysis of magnetic field variations was presented in Bailey
Figure 1. Substations in the Austrian transmission grid equipped with geomagnetically induced current measurement systems; DCC: Direct Current Compensation
System installed and measured; colored blocks with black numbers indicate the EURHOM earth layer model; WIC is the International Association of Geomagnetism
and Aeronomy (IAGA) code for the Conrad Observatory.
Figure 2. Measured current in A over the run times of installed measurement
systems from #01 to #08. (The #06 is not in use.) The measurement device #02
moved location in 2018, hence the two different colors.
ALBERT ET AL.
4 of 16
2.3. Geomagnetic Field Data
All measurements of the local geomagnetic field in Austria refer to those car-
ried out at the Conrad Observatory, a subterranean tunnel system located near
Muggendorf in Lower Austria. The observatory is an INTERMAGNET-qual-
ity (1 sample/sec) geomagnetic observatory (see https://intermagnet.org/ for
more details) with measurements starting in 2014. Only the x- and y-compo-
nents of the field (corresponding to the northward and eastward geomagnetic
field directions, respectively) are used in the analysis. Vertical (z) field varia-
tions, which generally do not contribute to geomagnetically induced currents
at the surface, are ignored (Boteler & Pirjola,2017). The individual horizon-
tal components are also combined into the horizontal magnetic field strength
component B that describes only the intensity in the 2D surface plane (where
). The geomagnetic variations dB/dt are expressed in change
in field strength per time-step, that is nT/min.
2.4. GIC Simulation
The geomagnetic storms that drive GICs lead to changes in the Earth magnet-
ic field (dB/dt) ranging from tens of nano Tesla (nT) up to several thousand
of nT per minute depending on the geomagnetic storm and the geographi-
cal location. As depicted in Figure3, the magnetic field propagates into the
electrically resistive subsurface, where it induces an electric field that can
be described by Faraday's Law of Induction. Under the assumption that the
field propagation can be treated as a plane-wave going into the Earth (Boteler
& Pirjola,2017), solving the differential Maxwell equation with the Euler
approach results in the following equations for the electric field in the north-
ward direction (Ex) and in the eastward direction (Ey):
. The variables ω, μ, σ, H, E j, and z are the angular frequency, the Earth permeability, the
electric conductivity, the magnetic field, the electric field, the imaginary unit, and the downward direction into
the Earth, respectively. Equations1 and2 imply that the electric field decreases with increasing penetration depth
z. Therefore, integrating the electric field along a closed loop, for example in the y- and z-direction (Figure3)
results in an electromotive force (emf) unequal to zero. This emf drives GICs in the loop formed by the power
grid connected to ground and the ground itself. The work done by emf on the electric charge can be measured as
a virtual electric potential around the loop. Further information on the electromagnetic field calculations can be
found in Simonyi(1971) and in Simpson(2005).
Two different approaches for GIC calculations are compared to measurements: the first uses the low frequency
current (LFC) simulator tool from Schachinger and Albert(2021), and the second approach is computed with a
fit of the geoelectric field components (Ex and Ey) to the GIC measurements (Equation3).
The LFC simulator uses the plane-wave method from Pirjola(1982) to calculate an electric field in the Earth's
surface. The plane-wave method is sufficient approximation for mid-latitude countries such as Austria, because
the magnetic field variations are usually dominated by distant magnetospheric currents and spatially smooth
mid-latitude ionospheric currents. The resistive Earth itself is model with 1D layers from the European Rho
Model (EURHOM, Ádám etal.,2012). Most of Austria can be described by two different resistivity models, one
for the alps in the west (EURHOM #55) and one for the flat lands in the east (EURHOM #39, Bailey etal.,2018).
A comparison of the different models, including the EURHOM models can be found in Bailey etal.(2018). In
contrast to the Lehtinen-Pirjola method (Lehtinen & Pirjola,1985), we use the nodal admittance matrix method,
Figure 3. Depiction of geomagnetic induction and geomagnetically induced
currents (GICs) in the power grid. ∂Hx/∂t are horizontal geomagnetic
variations in the northward direction, and Ey is the corresponding geoelectric
field induced in the eastward direction. The “GIC” loop shows the loop
formed between the ground and power lines, through which the GICs flow via
transformers. Layers of resistive material going into the Earth show how the
geoelectric field tends to lose intensity with increasing depth z.
ALBERT ET AL.
5 of 16
which is more common to use in the field of electrical engineering (Halbedl,2019). However, the two methods
are considered mathematically equivalent (Boteler etal.,2014). The power grid is modeled as a DC network with
voltage sources between the substation grounding and the resistive earth. The potential of all other earth refer-
ence points are changed relative to one overall reference point. This is only valid for uniform electric fields over
a certain area, however, it simplifies GIC calculation without losing much quality of the calculated GICs (Boteler
& Pirjola,1998; Halbedl,2019).
The second GIC modeling approach is based on the method described in Pulkkinen etal.(2007) and applied for
example in Torta etal.(2012). The geoelectric field is calculated using the plane-wave method using EURHOM
model #39, because, regardless of where the GIC measurements were made, the geoelectric field modeled using
#39 results in GICs that match the measurements well and better than any of the other models used, implying
this is a good approximation for most of the region. We expect that small-scale deviations do exist, although we
have yet to find any such locations. The GICs at different substations are calculated from the geoelectric field
components by applying a fit to the following equation:
where j is a specific substation measurement point in the power grid, and aj and bj are substation-specific co-
efficients that describe the contribution from each geoelectric field component and have the units A⋅km/V. The
substation coefficients need to be recalculated if the configuration of the power grid changes. The fit (Equa-
tion3) is applied to recent DC measurements—specifically a period in May 2021 with larger measured DC that
undoubtedly has geomagnetic sources to reduce input from other sources - and aj and bj can be determined using
the equations in Pulkkinen etal.(2007) or by carrying out a least-squares fit. Applying the method described
in Pulkkinen etal.(2007), the substation coefficients aj and bj incorporate uncertainties from the electric field
calculations, for example due to the limited Earth layer modeling.
An analysis is carried out of the entire DC measurement data set by first providing an overview of each meas-
urement station statistically and then by evaluating the different sources of DC both geomagnetic and man-made
source are identified.
3.1. Overview of Data
Table1 lists the transformer neutral point current data statistics. For the calculation of the statistics, outages and
measurement errors are removed. Switching events, which can exceed the measurement limit of ±25A, are not
removed. The maximum current refers to GIC events and not to switching events. These events are checked both
manually and with a peak detection algorithm.
Client# ndays raw ndays clean μ in A σ in A | GICmean |in A DCmax in A Date of DCmax UTC Sensitivity to Ex/Ey (% from Ex/% from Ey)
01 1,671 1,651 +0.13 0.33 0.26 −8.41 2021-05-12 12:21 49/51 (r=0.86)
02 479 427 −0.12 0.02 0.14 +0.83 2021-05-12 12:20 44/56 (r=0.84)
03 1,583 1,522 −0.20 0.03 0.21 −2.42 2017-09-08 23:02 45/55 (r=0.72)
04 1,564 1,512 +0.06 0.23 0.17 −4.57 2021-05-12 12:20 36/64 (r=0.85)
05 1,383 1,330 −0.07 0.21 0.14 −13.83 2021-05-12 12:20 06/94 (r=0.89)
07 579 492 −0.48 0.14 0.47 −2.72 2020-09-04 14:18 50/50 (r=0.85)
08 159 105 +0.24 0.76 0.56 +9.31 2021-05-12 12:48 76/24 (r=0.57)
Note. μ is the mean, σ the standard variation, and DCmax the maximum current measured to date caused by a geomagnetic event (i.e., not a data spike or transformer
switching events). The column "Sensitivity to Ex/Ey" provides the contribution of each geoelectric field component (in percentage) to the measured GICs according to
the GIC fit of modeled E to recent DC measurements. In brackets, r denotes the Pearson's correlation coefficient between the GIC fit and measurements (clean days).
Summary of the Data Properties (1-Sec Data) at Each Client
ALBERT ET AL.
6 of 16
We see that the stations experience very different levels of DC, with some having experienced GICs around 10A
and greater (DCmax in #01, #05 and #08) and others not even exceeding a maximum of 1A such as #02 during a
geomagnetic storm. Almost all of the peak GICs occurred during the May 2021 storm (which will be looked at
in more detail later in this work). There are also very different levels of noise (σ) ranging from extremely quiet in
#02 and #03, and very high levels in #01 and #08.
One question we can ask is how much of the measurements can be explained by geomagnetic sources? Included
in the last column is an estimate of the contribution of each geoelectric field component to the GICs measured
at each location. This was computed from a fit of the GIC measurements (interpolated to a sampling rate of one
minute) to the geoelectric field components modeled from geomagnetic field variations according to Equation3.
The values are for data from the week of 9th–16th May 2021, which included a geomagnetic storm and some of
the largest DC measurements to date. The r value in brackets gives the Pearson's correlation coefficient achieved
by the fit (For the stormy section alone on 12th May 2021, r ranges from 0.90 to 0.97 at all stations except for
#07 and #08, where it is 0.85 and 0.82, respectively). As the correlation is generally very high, we can deduce
that the signals seen in the stations can be explained by the geoelectric field variations for the most part. Some
are modeled better from E than others, often those with greater levels of measured DC (e.g., #01 and #05). The
lower r values for #08 are likely due to the very large noise level measured at that particular substation. In general,
the signal-to-noise ratio (SNR) and the GIC amplitude is related to the power grid topology (line orientation and
center/end node/substation). In addition to geomagnetic field variations, other transformer neutral point current
sources should be considered, in order to judge the actual GICs in the context of a risk analysis for the power grid
and its assets.
In the last column in Table1, most stations are somewhat balanced in contribution from both field directions, but
#05 stands out by having roughly 94% of the currents induced by the eastward geoelectric field component alone.
Client #05 is located at a transformer at the end of a long east-to-west 380kV transmission line and therefore
highly sensitive to eastward electric fields (Ey), which result from changes in the geomagnetic field in the x-di-
rection (dBx/dt), dominated by the magnetospheric ring currents, due to the location of Austria in the mid-latitude
region. The lines from the substation of #05 to Germany are 220kV lines. In addition, no auto-transformer is
used in the substation of #05.
Among the many signals seen in the DC measurements, there is a typical daily pattern, similar to the geomag-
netic field solar quiet variation. This quiet time variability in transformer DC measurements has recently been
investigated in detail in Kellerman etal.(2021). Figures4a–4d show the calculated mean values for each minute
of a day for magnetic field measurements and GIC measurement clients #01 and #05. For clients #01 and #05,
1,650days and 1,330days, respectively, were superimposed and the mean value of every minute was calculated.
The magnetic field measurements have fewer data gaps, therefore about 1,740days were used for the mean value
calculation. This reveals that there are recurring neutral point currents with different time periods at different
measurement points that are not caused by variations in the Earth's magnetic field. The excerpt of the fast Fourier
transform (FFT) analysis of the mean values of #01 and #05 in Figure4e shows the dominant frequency shares,
produced by other sources than geomagnetic field variations. For #01 these are sources with periods of 15min
and faster. These shares are also present in the frequency spectrum of #05, however, the main fast components
have duration times of 30min. For comparison purposes, #01 and #05 were normalized to their peak value before
performing the FFT.
The alleged time shift between the two curves in Figures4c and4d likely has two main causes: first, the measure-
ment systems are set to UTC time, but there is a difference in local time between the two clients of about 9min as
they are 250km apart. Second, the sensitivities to changes in the different components of the geomagnetic field
(resulting E-fields Ex and Ey have their maxima at different times) of the transformers in the two substations is
quite different, as one is more sensitive to variations in the x-direction than the other. Another interesting effect
that can be seen in Figures4c and4d is the small offset in both measurement clients. Although the measurement
systems were calibrated during installation, the overall mean value is not zero. This is likely caused by constant
DC currents unrelated to geomagnetic variations. For low frequencies the measurements from client #05 fit the
Bx field qualitatively very well.
ALBERT ET AL.
7 of 16
3.2. Noise Sources
As can be seen from Table1, there are varying levels of noise in the different measurements (mean μ and standard
deviation σ). Some of this can be explained by the location. Client #01 is located at the edge of the city of Vien-
na, where it is more likely that the noise is caused by earth-leakage currents from technical/man-made systems.
Client #05 is located ∼115km east of the city Munich and ∼50km east of a north-south railway transit corridor.
The increased noise could be caused by earth-leakage currents from the city of Munich, where there is also a DC
powered subway system operating, and by stray currents from the railway transit corridor/system.
Apart from these general noise sources, other sources of DC have been found in the measurements over the years.
So far we have been able to identify the Vienna DC powered subway system as one contributor to the transformer
neutral point currents. Another source of DC transformer neutral point currents are cathodic corrosion protection
systems of power plants and pipelines (Beltle etal.,2017). In addition, the galvanic coupling of the cathodic pro-
tection system and the transmission grid, hybrid AC/DC transmission lines are coupled through DC ion currents
(Pfeiffer etal.,2015). We are also currently investigating the effects of power electronics, such as the converters
from renewable energy resources (e.g., wind and photovoltaic, [Gertmar etal.,2005]), and the effects of trading
at the electrical energy market. Solar radiation can be excluded as a source (Albert etal.,2020).
Figure 4. Mean minute values of geomagnetic field measurements (a) Bx and (b) By, and DC measurement clients (c) #01
and (d) #05; (e) shows an excerpt from the normalized fast Fourier transform of the mean from #01 and #05.
Mean value of Bx
By in nT
Mean value of By
I in A
Mean value of Client #01
00:00:00 04:00:00 08:00:00 12:00:00 16:00:00 20:00:00
Time of day in UTC
I in A
Mean value of Client #05
10-4 10-3 10-2
Frequency in Hz
|I| in pu
normalized FFT of mean #01
normalized FFT of mean #05
ALBERT ET AL.
8 of 16
3.2.1. Vienna DC Subway System
As first described in Halbedl(2019), leakage currents of public transporta-
tion systems are measured in several measurement systems in Austria. This
was identified by analyzing the frequency spectrum of the currents, as well as
by comparing the operating hours of the Vienna subway with patterns in the
neutral point currents. During the nights from Sunday to Thursday, the sub-
way operation stops for a few hours (from roughly 00:30 a.m. till 05:00 a.m.),
and on the nights of Friday and Saturday, the subway stays in operation
through the night. The operating times can be seen clearly in Figure5a, which
shows the root-mean-square in the DC measurements over 8-day periods.
The currents clearly stay at a higher level during weekend nights rather than
dropping to around 0A, as they do on weekdays. As can be seen in the plot,
these stray currents contribute roughly 0.2A to the measurements during the
Due to the COVID-19 restrictions in public transportation, the Vienna sub-
way stopped its weekend nightly operations and also reduced the number of
trains in operation. This change in operating hours can be seen in Figure5b,
in which there are also lower neutral point currents during the weekend
nights. Lowes(2009) also identified unintended earth-leakage currents from
DC railways systems, which caused interference during geophysical surveys.
3.3. Recurrent Geomagnetic Activity
Having identified some sources of noise in the data, we now look more closely at the geomagnetic signals pres-
ent in the measurements, among which are recurrent geomagnetic activity. Geomagnetic activity has long been
known to recur roughly every 27days (Richardson etal.,2000; Tsurutani etal.,2006) due to the persistence of
high speed streams and corotating interactions regions (Alves etal.,2006) over more than one solar rotation or
Carrington rotation. This is evidenced by recurrent mild geomagnetic storms. A recent study by Gil etal.(2021)
has shown that this is also observable in power grid observations.
Figure6 shows the DC measurements across each Carrington rotation (roughly 27days long). To produce this
plot, each Carrington rotation was split into 100 time windows (each window one pixel), and the maximum abso-
lute DC or dH/dt (variation in the horizontal magnetic field strength) was found for each window, while the solar
wind speed, which clearly show corotating interaction regions in the solar wind, is plotted beneath. The recurrent
geomagnetic activity can be seen in both upper plots, although in the DC measurements it is not as pronounced as
in geomagnetic variations partly because there is a constant level of noise in the DC measurements, which some
of the geomagnetically induced currents can get lost in. Plot Figure6d shows the cross-correlations between each
time series at different time shifts. The highest correlations are between dH/dt and the solar wind, likely because
there is a better SNR. The 27-day recurrence from solar wind structures is clearly visible in the secondary peaks
at time shifts of 27 and 54days. The coherence between daily variations in dH/dt and measured GICs are also
3.4. Events During the Observation Period
Throughout the duration of the measurements and the progress of solar cycle 24 into cycle 25, we have observed
some minor and moderate geomagnetic storms. Here we present the measurements and a brief analysis of two
storms. For each storm, we also consider the cumulative GICs that would have been seen during that period by
calculating an additional parameter, GICsum.
3.4.1. Calculation of GIC-Sum
In the Austrian power grid, transformers in the 220 and 380kV levels are generally solidly grounded, without
any resistance between the transformer neutral point and the substation grounding. GICs can enter and exit the
transmission grid via the transformer neutral points. The transformers are designed to handle alternating mag-
netic flux in the magnetic transformer core, and direct magnetic flux can cause saturation of this magnetic core
Figure 5. Influence of Vienna DC subway leakage currents on transformer
neutral point currents as seen in: (a) normal operation, including over weekend
nights, where the subways stays in operation, and (b) during COVID-19
lockdown restrictions, when the subway stopped extended operation during
ALBERT ET AL.
9 of 16
material. Operating with a saturated transformer core has short-term and long-term effects on the transformer.
Short-term effects (scales of minutes) are half-cycle saturation leading to current and voltage distortion, which
are the reasons for the increased non-active power demand of the transformer. The increased non-active power
demand can cause undesired voltage drops and power system instabilities. If saturation in the transformer core
lasts over multiple hours or even days, the voltage and current distortion also causes transformer heating, which
can be considered as a long-term (scales of hours) effect. Short- and long-term effects of GICs on transformer
are also discussed in Gaunt etal.(2020). The transformer heating is due to the increased current and stray flux in
the metallic tank and transformer reinforcements. GICs with comparable short duration and high amplitudes as
well as GICs with comparable low amplitude and long duration can cause the transformer hot spot temperature
to increase above the acceptable design limits. A further increase in temperature causes a loss of insulation and
internal transformer failures.
Figure 6. Influence of recurrent geomagnetic activity from high speed streams and corotating interaction regions. (a) Measurements of geomagnetically induced
currents (GICs) at station #01 (from which we have the longest time series), (b) horizontal magnetic field variations (from the Conrad Observatory in Lower Austria),
(c) the measured solar wind speed showing corotating interaction regions, and (d) the cross-correlation between the time series for different shifts in time. Each variable
is plotted for each 27-day Carrington Rotation (x-axis) across all Carrington rotations from September 2016 till May 2021 (y-axis, time increasing from top to bottom).
Examples of recurring activity visible in both GIC and magnetic data that last longer than one rotation are marked with red circles in both plots. White spaces in both
plots are data gaps. The recurrent activity can be seen in the secondary cross correlation peaks at time shifts of 27 and 54days.
ALBERT ET AL.
10 of 16
In order to quantify and consider the afore-mentioned long-term effects, the GICsum value is calculated according
to Equation4. The GICsum value is the accumulated absolute current amplitude over a fixed time period with a
fixed sample rate, which in this case is a time period of one hour and a sampling rate of one second.
The data presented in Figures8 and 11 are a first attempt to quantify and take into accumulated GIC load on
transformer. For a further risk analysis, a guidance containing a threshold level where GICs starts to contribute
to an accumulated exposure and different alert levels should be provided. This guidance would be needed to be
adapted for different transformer designs.
3.4.2. 12 May 2021 Event
On 9th May 2021, a CME associated with a filament eruption was detected. This reached the Earth and became
geomagnetically effective on 12th May 2021. During this event, a maximum Kp value of 7 (GFZ German Re-
search Centre for Geoscience,2021) was reached between 12:00 and 15:00 UTC. At 14:00 UTC the Dst value
reached the minimum of −61nT (World Data Center for Geomagnetism, Kyoto, Japan, 2021). The maximum
amplitudes of currents measured in the Austrian transmission grid were 13.83 and 9.31A in clients #05 and client
#08, respectively. Figure7 shows the measured currents of three transformers and the corresponding results of
the LFC simulator tool and a fit of the geoelectric field to the data (GIC fit). For the clients #01 and #05 in a) and
b), the simulated currents fit the measured currents very well (Pearsons correlation coefficient r equals 0.84 and
0.96, respectively), however the results from the LFC simulator for client #08 in c) do not match well (r=0.22)
despite the fit matching the current well (r=0.81). This is probably caused by inaccurate grid data or uncertain-
ties in the Earth layer model.
In Figures8a, 8c and8e the 1hr GICsum value for the disturbed day May 12th is plotted, in b), d) and f) the 1hr
GICsum value for a comparable quiet period is plotted alongside. During the quiet period, the maximum Kp value
was 1-. Although the highest current amplitude was measured at client #05 (13.83A), the highest GICsum value
Figure 7. Measured currents during the May 2021 event, related simulation results of two calculation approaches and
corresponding Pearson correlation coefficients r on 12 May 2021. The selected transformers show the highest currents: (a)
client #01, (b) client #05 and (c) client #08 (the deviation of the results from the low frequency current simulator are probably
due to inaccurate grid data of this region).
I in A
Client #01 Measured
LFC Simulator r = 0.84
GIC fit r = 0.9
I in A
Client #05 Measured
LFC Simulator r = 0.96
GIC fit r = 0.95
11:3011:45 12:00 12:15 12:30 12:45
May 12, 2021
I in A
Client #08 Measured
LFC Simulator r = 0.22
GIC fit r = 0.81
ALBERT ET AL.
11 of 16
was reached at client #08. Clients #01 and #05 show a similar GICsum pattern
during both the geomagnetic disturbance and the quiet period. Both trans-
formers were exposed to comparable GICsum currents during a 1hr period.
Similar to the correlations done by Choi etal.(2015), Figure9 shows the cor-
relation between changes in the magnetic field and resulting GICs. As mag-
netic field measurements in nT/min are calculated for the x- and y-directions.
The related changes in measured currents at three measurement systems are
given in A/min. This reveals the sensitivity directions of these stations: client
#01 shows in a) and b) high correlations between changes in neutral point
current and changes of the geomagnetic field. Panels c) and d) show a higher
sensitivity of client #05 to changes in x-direction, which matches with cal-
culations in Halbedl(2019). The currents at client #08 are in general higher
than at #05, however, the changes caused by geomagnetic variations during
this event were smaller and a clustering can be seen in f), which also indicates
a lower sensitivity on changes in y-direction. The visual determined sensi-
tivities to geomagnetic fields in Figure9 also match with the sensitivities to
geoelectric fields in Table1.
3.4.3. 8-9 September 2017 Event
The September 2017 event was associated with a X9.3 flare from solar active
region (AR) 12673, which is regarded as the largest in solar cycle 24. A max-
imum Kp value of 8+ (GFZ German Research Centre for Geoscience,2021)
was reached between 12:00 and 15:00 UTC, although the maximum current
amplitude only reached 5.11A at client #01. During the September 2017 event, three measurement systems (#01,
#03, #04) were active. The current amplitude at client #03 is less than half the amplitude at client #01, therefore
only GICsum value of client #01 and #04 are plotted in Figure10. This is a particularly interesting comparison
because the two are located in the same substation in the 220 and 380kV
Again, the measured currents are compared with the two calculation ap-
proaches of LFC Simulator and GIC fit. Unfortunately, the September event
was measured with the first version of the measurement system, therefore
some of the peak values are cut off because of the −3.5A limit in the meas-
urement device, and saturated GIC measurements imply that the actual GICs
were larger. As the Dst value during the September 2017 event (−122nT)
was exactly twice the Dst value during the May 2021 event (−61nT), the
GIC during the September 2017 event is expected to be in the range of 25A.
In Figures11a and11c, the 1hr GICsum value for the most disturbed period
in September 2017 is plotted, and in b) and d) the 1hr GICsum value for a
comparable quiet period is plotted. During the quiet period the maximum
Kp value was 2+. Note that client #01 is a 380kV transformer neutral meas-
urement and #04 a 220kV measurement in the same power grid substation.
Due to the increased circuit resistance in the 220kV level, the transformer
neutral currents are usually lower than those in the 380kV transformer neu-
tral, and we see they are roughly half as large in #04 as they are in #01. In
Austria, the common practice is to have one transformer neutral per voltage
level and substation grounded. This is required to reliably detect faults in the
grid and to limit line-to-ground fault currents. Usually multiple transformers
are connected in parallel to ensure 1-n security and to increase the transmis-
sion capacity. In our case in the substation of #01 and #04, three 380/220kV
transformers are connected in parallel, where the 380kV neutral point of
transformer 1 is solidly ground and the 220kV neutral point of transformer 3
is solidly grounded. Therefore, the two measurement points are at 2 different
transformers. The other neutrals of the other transformers are not grounded.
Figure 8. Cumulative 1h GICsum of client #01 for May 2021 storm (a, c, e)
and quiet (b, d, f) period during May 2021.
a) active period; #01
May 12, 00:00 May 13, 00:00
b) quiet period; #01
May 05, 00:00May 06, 00:0
c) active period; #05
May 12, 00:00 May 13, 00:00
d) quiet period; #05
May 05, 00:00May 06, 00:0
e) active period; #08
May 12, 00:00 May 13, 00:00
f) quiet period; #08
May 05, 00:00May 06, 00:0
Figure 9. Correlation of dB/dt and the resulting change in neutral point
currents dI/dt during the 12th May 2021 event: (a and b) client #01, dI/dt
standard deviation σ=0.297 A/min, (c and d): client #05 σ=0.645 A/min, (e
and f): client #08 σ=0.461 A/min. The slope k of the black least squares fit
line is also shown.
dBx/dt in nT/min
#01 dI/dt in A/min
dBy/dt in nT/min
#01 dI/dt in A/min
dBx/dt in nT/min
#05 dI/dt in A/min
dBy/dt in nT/min
#05 dI/dt in A/min
dBx/dt in nT/min
#08 dI/dt in A/min
dBy/dt in nT/min
#08 dI/dt in A/min
k: 0.062 k: -0.138
ALBERT ET AL.
12 of 16
The highest current amplitude during the September 2017 event was measured at client #01 (5.11A), the highest
GICsum value was also reached at client #01 (1.9Ah). The transformer at client #04 in the same substation as
client #01 was exposed to a 1.1Ah during the same time period. During the quiet period with no geomagnetic
activity (Kp 0o), the transformer at client #01 and #04 were exposed to max. 0.33Ah.
The measurements covered a period of particularly low levels of geomagnetic activity throughout the end of solar
cycle 24 and the beginning of cycle 25. The maximum Kp of 8- was reached during the September 2017 storm.
During the last years of measurement, the geomagnetic activity was comparably low with very few periods of
increased activity. Nevertheless, with the recent storms we now have sufficient data to reliably calculate the GICs
in the Austrian power grid, which is unfortunately low for studying larger GIC events, however due to the relax-
ation of maximum measurement limits in the past year, we now have sufficient data to estimate GIC levels from
recent storms. A historical analysis has been carried out in another recent study (Bailey etal.,2022), giving an
estimate for the largest GICs during the 2003 Halloween storm that likely ranged from at least 30A up to 60A.
By splitting the GIC measurement data into time intervals matching Carrington rotation intervals, recurring solar
events could be identified in Figure6. High recurring changes in the magnetic field with time periods of 27days
can be seen magnetic field measurements as well as in neutral point measurements in the power grid. This un-
derlines the sensitivity of the power transmission grid, also to comparable small changes in the Earth magnetic
field. Small changes in the Earth's magnetic field can also have negative effects on the power grid. Low level
GICs cause an increased power loss in the system due to increased transformer windings losses and losses on the
overhead transmission lines (Forbes & St. Cyr,2010). In addition, the allowable transformer load can be reduced
by GICs (Girgis & Ko,1992).
Calculating mean values per minute for measurement periods reveals different noise sources with different fre-
quencies in the power grid. The noise sources do not only have specific frequency spectra, they also show daily
patterns, with changes every 15 or 30min. One of the noise sources was identified: by comparing time patterns
of GIC measurements, the influence of public transportation on transformer neutral point currents was discov-
ered. A change in the subways timetable due to COVID-19 restrictions, confirmed this theory. Typically, the
Vienna subways stops operation during weekday-night and continuous operation in weekend nights. However,
this changed in March 2020 due to COVID-19 restrictions. This effect on GIC measurements could only be
derived because of continuous and geographical distributed measurements. Further man-made sources of low
frequency currents in the transformer neutral point are currently investigated at the time of writing, but require
Figure 10. Measured currents during the September storm 2017 and related simulation results of two calculation approaches.
The selected transformers show the currents at two different neutral points in the same substation and the corresponding
Pearson's correlation coefficients: (a) client #01, (b) client #04; the drop in the measurement of client #01 after 12:30 UTC
and before 15:00 UTC is caused by maximum current measurable with the first version of the measurement system.
I in A
Client #01 Measured
LFC Simulator r = 0.86
GIC fit r = 0.84
12:0012:30 13:0013:30 14:0014:30 15:00 15:30 16:0
Sep 08, 2017
I in A
Client #04 Measured
LFC Simulator r = 0.89
GIC fit r = 0.88
ALBERT ET AL.
13 of 16
measurements with sample rates well above 1Hz. To address this, temporary
measurement recorders are being installed in various measurement locations.
The measurements are used for analyzing the influence of GICs on trans-
formers and calculating the theoretical influence on the power grid. Howev-
er, there are additional continuous long term measurements in power grids
available. Phasor measurements units (PMU) provide data from voltage, cur-
rents, angle and frequency measurements in the power grid. The correlation
of PMU data with high GICs will be done in future studies. Combining data
of multiple measurement types during geomagnetic active periods could also
reveal more direct influence on power grids, for example changes in power
flow or reactive power consumption. The combination could also reveal more
noise sources in the power grid DC measurements.
In addition to the current measurements, the GICsum value was calculated
to evaluate long-term effects of GICs on power transformers in addition to
short-term effects. The materials used in the transformers are designed for a
specific lifetime, and any increase of temperature above the design limits re-
duces the transformer lifetime (loss-of-life). Therefore, a thermal assessment
of the different transformer types in the fleet should be carried out in order to
determine the loss-of-life of transformers exposed to GICs. Regarding trans-
former overheating, Raith(2019) indicates that, for a specific transformer
design, a GICsum value of 1,000Ah, would be permissible without any over-
heating of the transformer. Note that the 1,000Ah could be reached with a transformer neutral point current of
60A and a duration of 1,000min or with a neutral point current of 10A and a duration of 6,000min (100hr,
4.17days). With this background, no transformer overheating is expected during the two presented GIC events
during the five year observation period, considering the same transformer design as in Raith(2019). But we
expect that increased moderate geomagnetic activity could cause a rise in transformer temperature. If an active
transformer cooling system is installed and not already in full operation, the cooling system could be used to
reduce the transformer operating temperature.
Besides uncertainties in geological structure, missing grid data is the main reason for differences between meas-
urement and simulation (see e.g., the model results for #08 in Figure7). A comparison of the magnetic field
variation over time with the three closest INTERMAGNET observatories shows a congruent pattern. Therefore,
the data from the single observatory (WIC) can be used in combination with the plane-wave method for GIC
studies in Austria. The Austrian power grid is a non-steady, varying infrastructure over the measurement period
of several years. This makes simulations difficult, as the current status of connections, outages or shutdowns is
not known for every day since 2016. The resistances of transformers and lines are well known, but the substation
grounding resistances are not available for all substations in Austria, and through experimentation with the sim-
ulation output versus measurements, these have been found to have a large effect on the modeled GICs (a change
of 0.1Ω lead to a maximum change of 32% or 2A in the specific transformer neutral). A dedicated measurement
campaign would be needed to gather more accurate grounding resistance data and account for this error source.
During the commission of substations the state-of-the-art is to measure the substation earthing impedance with
frequencies close to 50Hz. In order to improve the simulation, a substation earthing resistance measurement with
DC (0Hz) and/or with low-frequencies (below 50/60Hz nominal power system frequency).
Additionally, although we have data on the Austrian power grid, changes in neighboring countries are often
unknown but still have influence on our calculations. In the models we considered at least 2 substations into the
surrounding grids of DE, CH, CZ, SI, SK, CR and HU.
To conclude, we have presented an analysis of five years of DC measurements conducted with seven measurement
systems at multiple locations in the Austrian power grid. The maximum measured amplitude during the observa-
tion period was 13.83A during the geomagnetic storm on 2021 May 12th with a minimum Dst value of −61nT.
In addition to geomagnetic events and recurrent geomagnetic activity, we have identified the DC-powered public
Figure 11. Cumulative 1hr GICsum for September 2017 storm (a), (c) and
quiet (b), (d) period (data in during the quiet periods are below 0.02 Ah)
during September 2017.
a) active period; #01
Sep 07 Sep 08 Sep 09
b) quiet period; #01
Sep 25 Sep 26
c) active period; #04
Sep 07 Sep 08 Sep 09
d) quiet period; #04
Sep 25 Sep 26
ALBERT ET AL.
14 of 16
transportation system as a contributor to the grounded transformer neutral point currents. Other sources such as
power electronic devices from renewable energy resources are also under investigation. Location-dependent daily
recurring noise patterns have been detected in two sets of measurements and shown in Figures4c and4d, but the
sources have not been identified yet.
Correlations between changes in the magnetic field and changing neutral point currents during high geomagnet-
ic activity are shown in Figure9. In order to confirm the consistency of the measurement data, the correlation
between the magnetic field changes and the changing neutral point current was calculated. The results reveals
location-dependent GIC sensitivity to specific directions of geomagnetic field changes, which is also shown by
the GIC fit method in the last column of Table1.
Regarding the effects of GICs on transformer the 1hr GICsum value is calculated for the two presented events, re-
vealing no transformer temperature increase could be expected during the events. A winding temperature increase
is very unlikely, due to the values stated in Raith(2019), but cannot be excluded. Nevertheless, an individual
transformer thermal-fleet screening should be carried out to determine the GIC sensitivity among the transformer
types in the fleet.
The long term measurements in the Austrian power grid have already led to improvements in the simulation
accuracy for GICs in the Austrian power grid by initiating further studies (not yet published) in the field of sub-
station grounding calculation and sensitivity to various grid data. In addition to geomagnetic sources, we also
identified other sources of low frequency currents. This work supports transmission grid operators to maintain
and improve the grid availability and security. In the future, power grid assets and transformers endangered by
GICs can be identified and protected by mitigation actions. In addition, other systems that interact with the power
grid can be identified through continuous monitoring, providing reliable data to the system operation and utility
Data Availability Statement
The data this study is based on, namely the DC measurements in transformers in Austria, can not be openly
shared due to grid safety concerns, however the data may be shared in part for smaller studies, and any interested
party should contact the corresponding author for details. The geomagnetic field variations from the Conrad
Observatory (WIC) used to model the geoelectric field in Austria can be found under https://intermagnet.org/ or
by contacting the Observatory personnel. The LFC simulator used to model the GICs in the grid can be found
at https://github.com/P-Schachinger/LFC_simulator. The Dst used in this paper was provided by the WDC for
Geomagnetism, Kyoto (http://wdc.kugi.kyoto-u.ac.jp/wdc/Sec3.html). The Kp values used in this paper were
provided by the GZF German Research Centra for Geosciences, Potsdam (https://isdc.gfz-potsdam.de/kp-index/).
Ádám, A., Prácser, E., & Wesztergom, V. (2012). Estimation of the electric resistivity distribution (EURHOM) in the European lithosphere in the
frame of the eurisgic WP2 project. Acta Geodaetica et Geophysica Hungarica, 47(4), 377–387. https://doi.org/10.1556/AGeod.47.2012.4.1
Albert, D., Halbedl, T., Renner, H., Bailey, R. L., & Achleitner, G. (2019). Geomagnetically induced currents and space weather - A review of
current and future research in Austria. In 2019 54th International Universities Power engineering Conference (UPEC) (pp. 1–6). IEEE. https://
Albert, D., Schachinger, P., Renner, H., Hamberger, P., Klammer, F., & Achleitner, G. (2020). Field experience of small quasi DC bias on power
transformers A first classification of low-frequency current pattern and identification of sources. In Cigre Session 48. Paris (Vol. 137, pp.
Alves, M. V., Echer, E., & Gonzalez, W. D. (2006). Geoeffectiveness of corotating interaction regions as measured by Dst index. Journal of
Geophysical Research, 111(A7), A07S05. https://doi.org/10.1029/2005JA011379
Bailey, R. L., Halbedl, T., Schattauer, I., Achleitner, G., & Leonhardt, R. (2018). Validating GIC models with measurements in Austria: Evalua-
tion of accuracy and sensitivity to input parameters. Space Weather, 16(7), 887–902. https://doi.org/10.1029/2018SW001842
Bailey, R. L., Halbedl, T., Schattauer, I., Römer, A., Achleitner, G., Beggan, C. D., etal. (2017). Modelling geomagnetically induced currents in
midlatitude Central Europe using a thin-sheet approach. Annales Geophysicae, 35(3), 751–761. https://doi.org/10.5194/angeo-35-751-2017
Bailey, R. L., & Leonhardt, R. (2016). Automated detection of geomagnetic storms with heightened risk of GIC. Earth, Planets and Space, 68(1),
Bailey, R. L., Leonhardt, R., Möstl, C., Beggan, C., Reiss, M. A., Bhaskar, A., & Weiss, A. J. (2022). Forecasting GICs and geoelectric fields from
solar wind data using LSTMs: Application in Austria. Space Weather, 2021SW002907. https://doi.org/10.1002/essoar.10508699.1
Beltle, M., Schühle, M., & Tenbohlen, S. (2017). Galvanic coupling of direct currents in transmission grids and its effects on power transformers. In 20th
International Symposium on High Votlage Engineering. Institute of High Voltage and Power Transmission. Retrieved from https://www.research-
We would like to extend our thanks to
APG for giving us the opportunity to set
up and continuously operate the trans-
former neutral current measurements.
P. Schachinger and D. Albert thank the
TU Graz, APG and Siemens Energy for
funding. R. L. Bailey thanks the FFG and
ZAMG for funding. Supported by TU
Graz Open Access Publishing Fund.
ALBERT ET AL.
15 of 16
Bolduc, L. (2002). GIC observations and studies in the Hydro-Québec power system. Journal of Atmospheric and Solar-Terrestrial Physics,
64(16), 1793–1802. https://doi.org/10.1016/S1364-6826(02)00128-1
Boteler, D. H., Pirjola, R., Blais, C., & Foss, A. (2014). Development of a GIC simulator. In 2014 IEEE PES general meeting (pp. 1–5). IEEE.
Boteler, D. H., & Pirjola, R. J. (1998). Modelling geomagnetically induced currents produced by realistic and uniform electric fields. IEEE Trans-
actions on Power Delivery, 13(4), 1303–1308. https://doi.org/10.1109/61.714500
Boteler, D. H., & Pirjola, R. J. (2017). Modeling geomagnetically induced currents. Space Weather, 15(1), 258–276. https://doi.
Choi, K.-C., Park, M.-Y., Ryu, Y., Hong, Y., Yi, J.-H., Park, S.-W., & Kim, J.-H. (2015). Installation of induced current measurement systems
in substations and analysis of GIC data during geomagnetic storms. Journal of Astronomy and Space Sciences, 32(4), 427–434. https://doi.
Forbes, K. F., & St. Cyr, O. C. (2010). An anatomy of space weather’s electricity market impact: Case of the PJM power grid and the performance
of its 500 kV transformers. Space Weather, 8(9), S09004. https://doi.org/10.1029/2009SW000498
Fritsch, F. N., & Carlson, R. E. (1980). Monotone piecewise cubic interpolation. SIAM Journal on Numerical Analysis, 17(2), 238–246. https://
Gaunt, C. T., Cilliers, P. J., Heyns, M. J., Lotz, S. I., Chisepo, H. K., Adebayo, A. V., & Oyedokun, D. T. O. (2020). Calculations leading to voltage
stability and transformer assessment in the presence of geomagnetically induced currents. In Cigre Session 48 (Vol. C4). Paris. Retrieved from
Gertmar, L., Karlsson, P., & Samuelsson, O. (2005). On DC injection to AC grids from distributed generation. In European Conference on Power
Electronics and Applications, 2005 (p. 10). IEEE Service Center. https://doi.org/10.1109/EPE.2005.219420
GFZ German Research Centre for Geoscience. (2021). Geomagnetic observatories. Retrieved from https://www.gfz-potsdam.de/en/section/
Gil, A., Modzelewska, R., Wawrzaszek, A., Piekart, B., & Milosz, T. (2021). Solar rotation multiples in space-weather effects. Solar Physics,
296(8), 1–13. https://doi.org/10.1007/s11207-021-01873-7
Girgis, R., & Ko, C.-D. (1992). Calculation techniques and results of effects of GIC currents as applied to large power transformers. IEEE Trans-
actions on Power Delivery, 7(2), 699–705. https://doi.org/10.1109/61.127070
Halbedl, T. (2019). Low frequency neutral point currents on transformer in the Austrian power transmission network. (PhD Thesis). Graz Univer-
sity of Technology. Retrieved from https://diglib.tugraz.at/download.php?id=5cc8220f5d096&location=browsethesis
Halbedl, T., Renner, H., Bailey, R. L., Leonhardt, R., & Achleitner, G. (2016). Analysis of the impact of geomagnetic disturbances on the Austrian
transmission grid. In 19th Power Systems Computation Conference (pp. 1–5). IEEE. https://doi.org/10.1109/PSCC.2016.7540833
Halbedl, T., Renner, H., Sakulin, M., & Achleitner, G. (2014). Measurement and analysis of neutral point currents in a 400-kV-network. In 2014
Electric Power Quality and Supply Reliability Conference (PQ) (pp. 65–68). IEEE. https://doi.org/10.1109/PQ.2014.6866785
Kelbert, A. (2020). The role of global/regional earth conductivity models in natural geomagnetic hazard mitigation. Surveys in Geophysics, 41(1),
Kellerman, A. C., Mcgranaghan, R., Bortnik, J., Carter, B. A., Hughes, J., Arritt, R. F., etal. (2021). Geomagnetically induced currents at middle
latitudes: 1. Quiet–time variability. Space Weather. 19, e2021SW002729. https://doi.org/10.1029/2021SW002729
Lehtinen, M., & Pirjola, R. J. (1985). Currents produced in earthed conductor networks by geomagnetically-induced electric fields. Annales
Geophysicae, 3, 479–484.
Lowes, F. J. (2009). DC railways and the magnetic fields they produce—The geomagnetic context. Earth, Planets and Space, 61(8), i–xv. https://
Molinski, T. S. (2002). Why utilities respect geomagnetically induced currents. Journal of Atmospheric and Solar-Terrestrial Physics, 64(16),
Oughton, E. J., Skelton, A., Horne, R. B., Thomson, A. W. P., & Gaunt, C. T. (2017). Quantifying the daily economic impact of extreme space
weather due to failure in electricity transmission infrastructure. Space Weather, 15(1), 65–83. https://doi.org/10.1002/2016SW001491
Pfeiffer, M., Franck, C. M., & Schmutz, J. (2015). DC ion-currents in AC conductors in hybrid AC/DC transmission systems. In AC and DC
power transmission 2015. https://doi.org/10.1049/cp.2015.0022
Pirjola, R. (1982). Electromagnetic induction in the earth by a plane wave or by fields of line currents harmonic in time and space. (PhD Thesis).
Finnish Meteorological Institute. Retrieved from https://books.google.at/books?id=s33DwQEACAAJthesis
Pirjola, R. (1989). Geomagnetically induced currents in the Finnish 400 kV power transmission system. Physics of the Earth and Planetary
Interiors, 53(3–4), 214–220. https://doi.org/10.1016/0031-9201(89)90005-8
Price, P. R. (2002). Geomagnetically induced current effects on transformers. IEEE Transactions on Power Delivery, 17(4), 1002–1008. https://
Pulkkinen, A., Pirjola, R., & Viljanen, A. (2007). Determination of ground conductivity and system parameters for optimal modeling of geo-
magnetically induced current flow in technological systems. Earth, Planets and Space, 59(9), 999–1006. https://doi.org/10.1186/BF03352040
Raith, J. (2019). Risk assessment of power transformers under the influence of Geomagnetically Induced Currents (GIC). (Doctoral Thesis). TU
Graz. Retrieved from https://diglib.tugraz.at/risk-assessment-of-power-transformers-under-the-influence-of-geomagnetically-induced-cur-
Richardson, I. G., Cliver, E. W., & Cane, H. V. (2000). Sources of geomagnetic activity over the solar cycle: Relative importance of cor-
onal mass ejections, high-speed streams, and slow solar wind. Journal of Geophysical Research, 105(A8), 18203–18213. https://doi.
Rodger, C. J., Mac Manus, D. H., Dalzell, M., Thomson, A. W. P., Clarke, E., Petersen, T., & Divett, T. (2017). Long-term geomagnetically
induced current observations from New Zealand: Peak current estimates for extreme geomagnetic storms. Space Weather, 15(11), 1447–1460.
Schachinger, P., & Albert, D. (2021). LFC simulator. Institute of Electrical Power Systems. Retrieved from https://github.com/P-Schachinger/
Simonyi, K. (1971). Theoretische Elektrotechnik (4., überarbeitete und ergänzte Auflage ed., Vol. Band 20). VEB Deutscher Verlag der
Simpson, F. (2005). Practical magnetotellurics. Cambridge University Press. https://doi.org/10.1017/CBO9780511614095
Torta, J. M., Serrano, L., Regué, J. R., Sánchez, A. M., & Roldán, E. (2012). Geomagnetically induced currents in a power grid of northeastern
Spain. Space Weather, 10(6), S06002. https://doi.org/10.1029/2012SW000793
ALBERT ET AL.
16 of 16
Tsurutani, B. T., Gonzalez, W. D., Gonzalez, A. L. C., Guar nieri, F. L., Gopalswamy, N., Grande, M., etal. (2006). Corotating solar wind streams
and recurrent geomagnetic activity: A review. Journal of Geophysical Research, 111(A7), A07S01. https://doi.org/10.1029/2005JA011273
Zhang, J. J., Wang, C., Sun, T. R., Liu, C. M., & Wang, K. R. (2015). GIC due to storm sudden commencement in low-latitude high-voltage power
network in China: Observation and simulation. Space Weather, 13(10), 643–655. https://doi.org/10.1002/2015SW001263