Volume 20, Issue 4 e2021SW003005
Research Article
Open Access

Simulation of Geomagnetically Induced Currents in a Low-Latitude 500 kV Power Network During a Solar Superstorm

J. J. Zhang

Corresponding Author

J. J. Zhang

State Key Laboratory of Space Weather, National Space Science Center, Chinese Academy of Sciences, Beijing, China

University of Chinese Academy of Sciences, Beijing, China

Correspondence to:

J. J. Zhang,

[email protected]

Contribution: Conceptualization, Methodology, Writing - original draft, Writing - review & editing

Search for more papers by this author
Y. Q. Yu

Y. Q. Yu

School of Space and Environment, Space Science Institute, Beihang University, Beijing, China

Contribution: Software, Visualization

Search for more papers by this author
W. Q. Chen

W. Q. Chen

State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing, China

State Grid Hangzhou Yuhang Power Supply Company, Hangzhou, China

Contribution: Software, Visualization

Search for more papers by this author
C. Wang

C. Wang

State Key Laboratory of Space Weather, National Space Science Center, Chinese Academy of Sciences, Beijing, China

University of Chinese Academy of Sciences, Beijing, China

Contribution: Supervision

Search for more papers by this author
Y. D. Liu

Y. D. Liu

State Key Laboratory of Space Weather, National Space Science Center, Chinese Academy of Sciences, Beijing, China

University of Chinese Academy of Sciences, Beijing, China

Contribution: ​Investigation

Search for more papers by this author
C. M. Liu

C. M. Liu

State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing, China

Contribution: ​Investigation

Search for more papers by this author
L. G. Liu

L. G. Liu

State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing, China

Contribution: ​Investigation

Search for more papers by this author
First published: 23 March 2022
Citations: 1

Abstract

Geomagnetically induced currents (GICs) are one of the main manifestations through which space weather affects human technical facilities, and GICs constitute the final link in the solar wind-magnetosphere-ionosphere-ground interaction chain. Therefore, understanding the responses of power grids to solar superstorms is crucial for space weather research and emergency preparedness purposes. In this work, we combine the Space Weather Modeling Framework with a ground conductivity model and a model of the Chinese Guangdong 500 kV power grid to investigate the response of the whole power grid if the 23 July 2012 solar superstorm had struck the Earth. The maximum |GICs| produced in the power grid by this event reach approximately 400 A, which is more than thrice those measured during a strong magnetic storm with Kp = 8. Based on statistical analyses of the strength and duration of large GICs at 54 substations and a comparison with the GICs responsible for the Hydro-Québec power outage on 13 March 1989, we stipulate that the risk of GICs in the Guangdong 500 kV power grid is very high. The simulation results also reveal that field-aligned currents can play an important role in producing GICs in middle- and low-latitude power grids during solar superstorms. This finding provides crucial insight for understanding the factors that generate strong GICs at middle and low latitudes.

Key Points

  • We built a model to simulate geomagnetically induced currents (GICs) in a power grid in China during a solar superstorm

  • GICs can reach hundreds of amperes in a low-latitude power network during a solar superstorm

  • Field-aligned currents play an important role in producing GICs within low-latitude power networks during solar superstorms

Plain Language Summary

The Sun is not a particularly peaceful entity, it can abruptly launch powerful radiation and plasma into space. If such a coronal mass ejection (also called a solar storm) struck the Earth, there would be serious consequences for humanity. Specifically, solar storms can generate geomagnetically induced currents (GICs) in high-voltage power transmission systems, which can damage transformers and even cause the whole system to collapse. How would a power grid respond to an extreme solar storm? Furthermore, how strong can the associated GICs be, and how serious are the consequences? To address these uncertainties, we built a model of the corresponding interaction chain based on physical principles. The model was driven by the solar wind conditions of a solar superstorm that occurred on 23 July 2012 and was observed by a satellite. We showed that even a low-latitude power grid may collapse under the impact of a solar superstorm. Consequently, without sufficient preparation, the consequences of a solar superstorm for power grids worldwide could be devastating.

1 Introduction

Coronal mass ejections (CMEs) directed toward the Earth produce the most severe geomagnetic storms in the Earth's atmosphere. The highly variable magnetosphere-ionosphere current system produced by a CME can induce geoelectric fields in the ground, resulting in the generation of low-frequency geomagnetically induced currents (GICs) in conductor based technological infrastructure systems in the ground, such as power grids, oil and gas pipelines, and high-speed railway systems. The flow of large GICs through transformer windings can saturate the core and cause the transformer to malfunction (with the possibility of permanent damage) and can even cause the whole system to collapse (Pirjola, 2002). The corresponding increase in soil pile potentials can destroy the cathodic protection, accelerating the corrosion rate of pipelines (Boteler, 2000). Moreover, GICs flowing through the track circuits of railways may cause anomalies in signaling and control systems (L. Liu et al., 2016).

The most serious power grid incident on record attributable to GICs was the Hydro-Québec power outage during the March 1989 geomagnetic storm. Millions of Hydro-Québec customers suffered from the cold of winter were without power for 9 hr during the event. It was also reported that the GICs produced during this event damaged some extra high-voltage transformers and other grid components in Canada and the United States (Jonas & McCarron, 2015). The most intense geomagnetic storm observed in the past few hundred years is referred to as the Carrington event, which occurred on 1–2 September 1859. Undoubtedly, the economic and social well-being of people today is significantly dependent on technologies powered by the electrical infrastructure. In many cases, new electrical technologies rely on a stable supply of electrical power. As a result, large-scale blackouts can have serious economic impacts even if power is restored within a few hours. It follows, then, that the consequences of a Carrington-level event striking Earth unprepared could be catastrophic. In addition to having direct adverse effects on power networks and other infrastructure components, a sustained power outage might lead to a breakdown of the supply chain and even social instability. Global insurance market Lloyd's estimated that a Carrington-level space weather event could cause at least 20 million people in North America to lose power for up to 2 years and the time to replace all the damaged systems was estimated at between 4 and 10 years (Jonas & McCarron, 2015). Of course, these are only suppositions. Accordingly, the seriousness of the adverse effects of a solar superstorm on present-day power networks and other technological infrastructures demands further study (Riley et al., 2018).

Solar superstorms are unusual and rare solar eruption events. They have been defined by many works (e.g., Y. D. Liu et al., 2019; Riley et al., 2018, and others). Because solar superstorms occur very infrequently, the major challenge to understanding the magnetosphere-ionosphere dynamics and the effects of solar superstorms in the Earth's atmosphere on ground infrastructures is the lack of observation data (A. Ridley et al., 2006). On 23 July 2012, NASA's Solar Terrestrial Relations Observatory (STEREO)—A mission observed a giant, superfast interplanetary coronal mass ejection (ICME). The average transit speed was estimated to be 2,780 km/s (Russell et al., 2013). Fortunately, this event was not directed at the Earth. Nevertheless, the 2012 event aroused considerable unease within the space science community. Y. D. Liu et al. (2014) showed that the superstorm was formed by the in-transit interaction between two successive CMEs; they defined this CME as an “ICME-in-sheath” event, the preceding ejecta of which are further squeezed both by the host ejecta and by shockwave-induced compression. According to their analysis, the magnitudes of the magnetic field, speed, and plasma density of this ICME are the highest on record at 1 au (Y. D. Liu et al., 2020). Considering the consequences of a solar superstorm this powerful striking the Earth, Baker et al. (2013) estimated that the “worst-case scenario” would have involved an extreme geomagnetic storm, potentially significantly larger than the famous Carrington event of 1859. Therefore, they suggested that this event be adopted as the prototypical extreme space weather event for emergency preparedness purposes.

Some researches have sought to estimate what geoelectric fields or GICs would have been produced if the 23 July 2012 solar superstorm had struck the Earth. Ngwira et al. (2013) calculated the induced geoelectric fields at specific INTERMAGNET magnetometer sites by using global geomagnetic field perturbations generated under the Space Weather Modeling Framework (SWMF) as if the 23 July 2012 solar superstorm had been directed at the Earth. In their calculation, the resistive-end Québec ground conductivity model was applied at all INTERMAGNET sites. The resulting maximum induced geoelectric field exceeded 14 V/km in the auroral region, while the geoelectric field was lower than 1 V/km in low-latitude regions. However, they did not calculate the corresponding GICs in their work. In addition to geomagnetic perturbations, the local ground conductivity is important for computing GICs (Love et al., 2016). J. J. Zhang et al. (2016) extended the simulation effect to include local GICs flowing in a power grid and employed the local ground conductivity model of a low-latitude substation area, namely, the Chinese Shuanglong substation. They simulated the GICs flowing through this substation under the effect of the extreme interplanetary shock (IP shock) in the early stage of the 23 July 2012 event and showed that the maximum geoelectric field at the substation reached approximately 2 V/km, which was much stronger than the low-latitude geoelectric fields calculated by Ngwira et al. (2013). The resulting GICs at the Shuanglong substation was 12.6 A, which flowed from the earth to the neutral point of the transformer. Nevertheless, in their work, J. J. Zhang et al. (2016) focused on the GIC at one substation induced by the extreme IP shock of the 23 July 2012 event. The variation in solar wind conditions after such an extreme IP shock is incredibly complex, as is the response of the Earth. So far, the distribution and evolution of GICs in a large-scale power network comprising many substations and transformer lines throughout a solar superstorm event have yet to be studied.

On the other hand, the solar wind-magnetosphere-ionosphere coupling responsible for inducing large GICs is not well understood. Accordingly, some studies have been carried out on this topic in recent years. For instance, Adebesin et al. (2016) investigated the interplanetary and magnetospheric causes of extreme dB/dt in equatorial regions. They concluded that these rapid equatorial enhancement in dB/dt are caused by electric fields originating from magnetospheric currents. Likewise, Apatenkov et al. (2020) discovered that auroral omega bands can result in large GICs in auroral regions. Furthermore, upon assessing intense local dB/dt variations across North America, Ngwira et al. (2018) considered the magnetospheric waves produced during substorms to be the major driver of local dB/dt extrema. More recently, Wei et al. (2021) found that near-Earth bursty bulk flows can drive intense dB/dt variations in high-latitude regions.

In this study, we combine the SWMF with a ground conductivity model and a large-scale low-latitude power grid model of southeastern coastal China to simulate the response of the whole power grid if the 23 July 2012 solar superstorm had struck the Earth, and we assess the corresponding GIC risk of the power grid. We further investigate the contributions of different current systems to the large GICs in low-latitude regions during the extreme space weather event by taking advantage of the SWMF. The remainder of the paper is structured as follows. In Section 2, we introduce the methodology and input data sources. In Section 3, we show the simulation results and analyze the GIC risk of the power grid. In Section 4, we show the contributions of different current systems to the large GICs within the power grid. Finally, we discuss and summarize our findings in Section 5 and 6, respectively.

2 Methodology and Data

The calculation of GICs in a technological system can be divided into two parts: a geophysical part, which determines the horizontal geoelectric field, and an engineering part, which computes the GICs produced by the electric field (Pirjola, 2002). The input of the geophysical part includes information about variations in the geomagnetic field near the surface and the Earth's conductivity structure. In this study, the geomagnetic field perturbations at the Earth's surface calculated by the SWMF model are applied to determine the induced geoelectric field. Then, the induced geoelectric field provided by the geophysical part and the power network configuration and resistances are adopted as the input of the engineering part.

2.1 The SWMF Model and Driving Conditions

The SWMF model developed at the University of Michigan is used in this study to calculate the geomagnetic field perturbations at the Earths surface. Three domains, namely, the Block-Adaptive Tree Solar wind Roe-type Upwind Scheme (BATS-R-US) code (Powell et al., 1999), the Rice Convection Model (RCM; Toffoletto et al., 2003), and the Ridley Ionospheric Model (RIM; A. J. Ridley et al., 2004), are coupled in this study. The numerical domain is taken to be −224RE < x < 32RE, −128RE < y < 128RE, and −128RE < z < 128RE with a minimum grid spacing of 1/8RE. The inner boundary is set at 2.5RE centered on the Earth, where RE is the Earth's radius. The geomagnetic field perturbations are computed by integrating over all the current systems in space, including the magnetospheric, ionospheric, and gap region current systems, using Biot-Savart's law (e.g., Yu et al., 2010; J. J. Zhang et al., 2020, and others). The model is driven by the solar wind and interplanetary magnetic field (IMF) measured by the Plasma and Suprathermal Ion Composition (PLASTIC) and In situ Measurements of Particles and CME Transients (IMPACT) instruments onboard STEREO-A. The solar wind and IMF parameters presented in Y. D. Liu et al. (2014) are used here. Unfortunately, measurements of the proton density and temperature during the 23 July 2012 solar superstorm are largely missing because of the limitations of the detectors. Where data are missing, the electron densities with an energy greater than 45 eV are multiplied by a factor of 5 as a proxy for the proton density (Y. D. Liu et al., 2014). Y. D. Liu et al. (2014) showed that this proxy matches the proton density well when proton data are available during the time interval. Similarly, the plasma temperature is calculated from the observed solar wind speed. The temporal resolutions of the solar wind velocity and IMF data are 5 min and 20 s, respectively, and all the upstream data are processed to a resolution of 1 min for the simulation. From top to bottom, Figure 1 shows the three IMF components, solar wind speed, solar wind density, and temperature from 12:00 UT on 23 July 2012 to 12:00 UT on 24 July 2012. The blue dashed line at 20:55 UT on 23 July denotes the time of the IP shock, and the red dashed line at 00:45 UT on 24 July denotes the time when IMF Bz turned from strongly northward to southward. As shown in Figure 1, IMF Bz was strongly disturbed after the arrival of the IP shock. The IMF maintained its strong northward direction for approximately 2 hr after a two-hour disturbance period, after which it rapidly turned very strongly southward and reached a minimum of approximately −44 nT at 00:58 UT on 24 July.

Details are in the caption following the image

Interplanetary magnetic field (IMF) and solar wind parameters observed by STEREO-A from 12:00 UT on 23 July to 12:00 UT on 24 July 2012. From top to bottom, the panels show the three IMF components (Bx, By, and Bz), solar wind speed Vsw, plasma density N, and temperature T. The blue dashed line at 20:55 UT on 23 July denotes the time of the interplanetary shock, and the red dashed line at 00:45 UT on 24 July denotes the time when IMF Bz turned from strongly northward to southward.

2.2 The Induced Geoelectric Field

In this study, the GICs in the Guangdong 500 kV power grid are simulated and analyzed. The one-dimensional (1-D) layered Earth conductivity model of Guangdong Province in China that was built on the basis of magnetotelluric measurements (Li et al., 1987) is adopted to calculate the induced geoelectric field. For this conductivity model, the conductivity depends only on depth, and conductivity of each layer is assumed to be laterally uniform. The conductivity of each layer is listed in Table 1. Based on the plane wave assumption (Cagniard, 1953), the horizontal components of the induced geoelectric field Ex,y can be calculated from the perpendicular geomagnetic perturbations By,x and the 1-D Earth conductivity model in the frequency domain (Viljanen et al., 2004):
urn:x-wiley:15427390:media:swe21292:swe21292-math-0005(1)
urn:x-wiley:15427390:media:swe21292:swe21292-math-0006(2)
where the subscripts x and y denote the north and east geographic directions, respectively; μ0 is the vacuum permeability; and Z is the surface impedance, which depends on the Earth's conductivity structure. The geomagnetic perturbations at the Zhaoqing geomagnetic observatory (geographic latitude ∼23.01°, geographic longitude ∼113.44°) with a temporal resolution of 1 min calculated from the SWMF model are used to represent the geomagnetic perturbations in the whole area the power grid. Then, these perturbations are applied to determine the induced geoelectric field in this area.
Table 1. The Earth's Conductivity Model for Guangdong Province
Depth (km) Conductivity (S/m)
0 − 100 0.0005
100 − 150 0.01
150 − 200 0.1
>200 0.01

2.3 The Power Grid Model

Unlike electric power transmission systems, which employ alternating current at frequencies of 50–60 Hz, GICs can be treated as direct currents. The method developed by Lehtinen and Pirjola (1985) (hereafter denoted the LP method) is applied to model the GICs in the Guangdong 500 kV power grid which includes 54 substations and 62 transmission lines. The LP method considers an arbitrary discretely grounded network impacted by an external field and represents the electric field as a voltage source in circuit analysis. The voltages driving the GICs in the power grid can be calculated by integrating the geoelectric field along the transmission lines:
urn:x-wiley:15427390:media:swe21292:swe21292-math-0007(3)
where A and B represent the two ends of a given transmission line. Therefore, the GICs can be calculated via standard circuit theory based on Ohm's and Kirchhoff's laws. This approach has been widely adopted to model the GICs in power grids (e.g., Pirjola, 2007; Wang et al., 2021; Zheng et al., 2012, and others). In this method, the GICs at each subsection can be computed by solving the following matrix equation:
urn:x-wiley:15427390:media:swe21292:swe21292-math-0008(4)
where I is an N × 1 matrix representing the GICs flowing through each substation, N is the number of substations, 1 is an N × N identity matrix, Y is the N × N admittance matrix, Z is the N × N grounding impedance matrix, and J is the N × 1 grounding current matrix under ideal grounding conditions.

The equivalent grounding resistances for the 54 substations in the Guangdong 500 kV power grid are listed in Table 2, and the resistances and lengths of the 62 transmission lines are given in Table S1 of Supporting Information S1.

Table 2. The Equivalent Parameters of the Grid Nodes Within the Guangdong 500 kV Power Grid
No. Name Grounding resistances (Ω) No. Name Grounding resistances (Ω)
1 Hezhou 1.7 2 Wuzhou 1.7
3 Qujiang 1.6 4 Xianlingshan 1.7
5 Huadu 1.6 6 Beijiao 1.6
7 Zengcheng 1.6 8 Boluo 1.6
9 Luodong 1.4 10 Shuixiang 1.4
11 Hengli 1.3 12 Guancheng 1.3
13 Dongguan 1.5 14 Pengcheng 1.3
15 Shenzhen 1.5 16 Kunpeng 1.6
17 Guangnan 1.4 18 Cangjiang 1.7
19 Xijiang 1.4 20 Yandu 1.6
21 Shunde 1.4 22 Xiangshan 1.4
23 Guishan 1.6 24 Guoan 1.6
25 Wuyi 1.6 26 Dieling 1.6
27 Maoming 1.6 28 Gangcheng 1.6
29 Huizhou 1.5 30 Shangzhai 1.4
31 Maohu 1.6 32 Jiaying 1.6
33 Rongjiang 1.6 34 Shantou 1.6
35 Hanjiang 1.6 36 Pingshi 1.6
37 Guangxu 1.4 38 Huixu 1.4
39 Aoliyou 1.6 40 Yangxi 1.6
41 Tonggu 1.3 42 Zhuhai 1.6
43 Ling'ao 1.3 44 Honghaiwan 1.6
45 Haimen 1.5 46 Tuolin 1.6
47 Heshuyuan 1.6 48 Shajiao 1.2
49 Jiangmen 1.6 50 Fushan 1.7
51 Echeng 1.3 52 Suidong 1.0
53 Zhaoqing 1.3 54 Baoan 1.3

3 Simulation Results and Analysis

3.1 Simulation Results

Figure 2 shows a cross section of the magnetospheric response to the simulated solar superstorm in the Y = 0 plane at 01:00 UT on 24 July 2012. This period begins immediately after IMF Bz has turned from strongly northward to southward. The average value of the magnetopause standoff distance is approximately 11 RE during normal activity (Fairfield, 1971). Figure 2 shows the enhanced current density, J, and the magnetopause standoff distance, which is severely compressed to within 4RE. The magnetopause standoff distance is defined by the distance from the center of the Earth to the last closed field line of the magnetosphere measured along the geocentric solar magnetic line connecting the Sun and the Earth. Y. D. Liu et al. (2020) empirically estimated the upper limit of the magnetopause standoff distance to be approximately 3.3RE under the effect of a solar superstorm. In addition, Figure 3 shows the distribution of the field-aligned currents (FACs) in the Northern Hemisphere during the same period as Figure 2, where red represents upward-flowing current and blue representing downward-flowing current. The figure shows that the Region 1 FACs are dominant during this period.

Details are in the caption following the image

Example simulated cross section of the magnetosphere at 01:00 UT on 24 July 2012 from the Space Weather Modeling Framework model. The plot shows the current density, J, in color, and the magnetic field lines are plotted in white.

Details are in the caption following the image

Field-aligned currents in the Northern Hemisphere at 01:00 UT on 24 July 2012. Red represents upward-flowing current, while blue represents downward-flowing current. The geomagnetic pole is at the center, and the outer circle represents 55° geomagnetic latitude.

Figure 4 shows the SWMF-derived geomagnetic perturbations (Bx and By), the horizontal components of the induced geoelectric field (Ex and Ey) and the GICs at the Hezhou substation (24.51°N, 111.51°E) from 12:00 UT on 23 July to 12:00 UT on 24 July. The Hezhou substation is denoted as substation No. 1 in Table 2 and substation S1 in Figure 5. Figure 4 reveals that Bx experienced multiple negative excursions, with the first excursion on 23 July reaching a minimum of −246.27 nT at 22:47 UT, which is deviated from 0 nT at the start of the storm. The second main excursion on 24 July reach a minimum of −498.14 nT at approximately 01:00 UT on 24 July, and the third excursion reach a minimum of −532.95 nT at 04:26 UT on 24 July. In contrast, By featured only two negative excursions, with the first excursion on 23 July reaching a minimum of −329.74 nT at 22:44 UT and the second excursion on 24 July reaching a minimum of −278.41 nT at 01:03 UT. Then, there was a long and slow recovery period. Note that all minima represent deviations from 0 nT at the start of the storm. The maximum Ex exceeded 2.23 V/km at 22:48 UT on 23 July, while the minimum Ex reached −1.67 V/km at 01:00 UT on 24 July; likewise, the maximum Ey was 2.07 V/km at 00:58 UT on 24 July, while the minimum Ey was −2.33 V/km at approximately 21:10 UT on 23 July. The |Ex| and |Ey| maxima both exceeded 2.0 V/km, which is higher than the value reported by Ngwira et al. (2013) in a simulation of the same event. According to Figure 4 in Ngwira et al. (2013), the maximum induced geoelectric field simulated in the latitudinal zone of 20°–30° was less than 1.0 V/km. The difference between the values obtained in these two works must be due to the geomagnetic perturbations at different geomagnetic stations and the fact that a more regionally accurate local ground conductivity model is used in this study. The GICs at the Hezhou substation reached a maximum of 308.61 A at 22:49 UT on 23 July and its a minimum of −390.47 A at 00:58 UT on 24 July. Positive GICs indicate currents flowing from the ground to the neutral point of the transformer, while negative GICs represent currents flowing from the neutral point of the transformer to the ground.

Details are in the caption following the image

Space Weather Modeling Framework-derived geomagnetic perturbations (Bx and By), horizontal components of the induced geoelectric field (Ex and Ey) and geomagnetically induced currents at the Hezhou substation (24.51°N, 111.51°E) from 12:00 UT on 23 July to 12:00 UT on 24 July 2012.

Details are in the caption following the image

Geomagnetically induced currents at 54 substations within the Guangdong 500 kV power grid at 00:58 UT on 24 July 2012. The size of the disk is scaled with the magnitude of the geomagnetically induced currents. Red disks denote currents flowing from the ground to the neutral point of the transformer, while black disks denote currents flowing from the neutral point of the transformer to the ground.

The GICs calculated at the Hezhou substation shown in Figure 4 were the strongest GICs in the entire Guangdong 500 kV power grid during this event. Figure 5 illustrates the GICs at all 54 substations within the Guangdong 500 kV power grid at 00:58 UT on 24 July 2012, where the location of the Hezhou substation (substation No. 1) is labeled S1. In Figure 5, the size of the disk is scaled with the magnitude of the GICs; in addition, red disks denote currents flowing from the ground to the neutral point of the transformer, while black disks denote currents flowing from the neutral point of the transformer to the ground. The figure shows that the GICs vary greatly among the substations. Considering that the geomagnetic perturbations in the area surrounding the power grid are the same at all of the substations in the simulation and that the conductivity is assumed to be laterally uniform, we attribute the considerable difference in GICs among the substations mainly to the power grid characteristics, including the resistances and lengths of the transmission lines, the resistances and structure of the transformers, and the power grid topology. At 00:58 UT on 24 July 2012, the GICs flowing through 15 of the 54 substations exceeded 100 A.

3.2 Analysis

Previous research has shown that when the GIC of a single-phase core transformer reaches 30 A, the oil tank locally overheats; furthermore, when the GIC reaches 45 A, the iron core clamp locally overheats (B. Zhang et al., 20102011). Figure 6 sorts the absolute maximum GIC for each substation within the Guangdong 500 kV power grid from large to small during the 23 July 2012 solar superstorm. The pie chart shows the percentage of substations in each of five levels classified according to the maximum |GIC|: |GIC| < 50 A (26%), 50 A ≤ |GIC| < 100 A (22%), 100 A ≤ |GIC| < 200 A (33%), 200 A ≤ |GIC| < 300 A (13%), and |GIC| ≥ 300 A (6%). In other words, 74% of the substations experienced a maximum |GIC| exceeding 50 A, indicating that very high GICs flowed through the Guangdong 500 kV power grid during this event. The maximum |GIC| recorded at the Ling'ao nuclear power plant (S43 in Figure 5) during this solar superstorm is 277.66 A, which is more than thrice the historical maximum of 75.5 A measured at this substation during the strong magnetic storm (Kp = 8) that occurred on 9–10 November 2004 (C. M. Liu et al., 2009).

Details are in the caption following the image

Maximum |GIC| at each of the 54 substations within the Guangdong 500 kV power grid sorted from large to small during the 23 July 2012 solar superstorm. The percentage of substations in each of five levels classified by the maximum |GIC|: |GIC| < 50 A, 50 A ≤ |GIC| < 100 A, 100 A ≤ |GIC| < 200 A, 200 A ≤ |GIC| < 300 A, and |GIC| ≥ 300 A.

In addition to the maximum |GIC|, the GIC duration is another important factor that can damage transformers. The longer the GIC duration is, the higher the risk of a transformer overheating and burning. The total duration of |GICs| > 30 A at each substation during the whole event is plotted in Figure 7, and the pie chart shows the percentage of substations classified by duration into four categories: T ≥ 120 min, 120 min < T ≤ 60 min, 60 min < T ≤ 30 min, and T < 30 min. As shown in this figure, 55% of substations exhibited |GICs| > 30 A for a cumulative duration exceeding 30 min.

Details are in the caption following the image

Total duration of |GIC| > 30 A at each of the 54 substations within the Guangdong 500 kV power grid sorted from large to small during the 23 July 2012 solar superstorm. The percentage of substation is classified by duration into four categories: T ≥ 120 min, 120 min < T ≤ 60 min, 60 min < T ≤ 30 min, and T < 30 min.

The Hydro-Québec power outage that occurred on 13 March 1989 at a storm level of Kp = 9 was the most serious blackout caused by GICs in history. The direct cause of the Hydro-Québec power outage was successive tripping of seven static volt ampere reactive compensators (SVCs) in the La Grande system by GICs; the loss of these SVCs resulted in insufficient reactive power within the grid, which eventually caused a power outage. Boteler (2001) estimated the in the Hydro-Québec 735 kV system during the March 1989 storm as 234 A/phase at the Arnaud substation, 177 A/phase at Chénier, 150 A/phase at LG2, 129 A/phase at Churchill Falls, and 90 A/phase at four other substations. According to Figures 6 and 7, the GICs flowing through the Guangdong 500 kV power grid during the 23 July 2012 solar superstorm are comparable to or even larger than those in the Hydro-Québec 735 kV power grid during the March 1989 storm. This indicates that when a solar superstorm strikes the Earth, even the power grids in low-latitude areas are at a very high risk of breaking down.

4 Contribution of Different Current Systems to GICs

At high latitudes, GICs are typically produced by the auroral electrojets (Pulkkinen et al., 2003; Viljanen, 1997), whereas magnetospheric and ionospheric currents (such as the magnetopause current, ring currents, and the equatorial electrojet [EEJ]) are considered the possible sources of large GICs at low latitudes (Carter et al., 2015; Fiori et al., 2014; Kappenman, 2003; Marshall et al., 2012; Pulkkinen et al., 2012). Generally, ring currents are considered the main current system responsible for the large GICs that arise in middle- and low-latitude regions during geomagnetic storms. However, this hypothesis has not been confirmed. Moreover, the contributions of FACs to the GICs during geomagnetic storms in these regions may have been overlooked. By taking advantage of the SWMF model, the geomagnetic perturbations generated by different current systems can be differentiated, allowing the contributions of different current systems to the large GICs in low-latitude regions during geomagnetic storms to be quantified.

In the SWMF simulation, the modeled geomagnetic perturbation components are the sums of the perturbations caused by magnetospheric currents (GM), the FACs in the gap region (FAC), and the ionospheric Hall (Hall) and Pedersen (Pedersen) currents at high (>50°) latitudes (Yu et al., 2010). We plot the geomagnetic perturbations at the Hezhou substation (S1 in Figure 5) produced by different current systems in Figure 8, showing (from top to bottom) the x and y components of the geomagnetic perturbations, the derivatives of the x and y components of the geomagnetic perturbations, and the simulated GICs. The gray dashed line at 00:45 UT on 24 July denotes the time when IMF Bz turned from strongly northward to southward. It is generally believed that ring currents are the main contributor to Bx during geomagnetic storms at low latitudes, while FACs are believed to be the main contributor to By. Here, the ring currents are included with the GM currents.

Details are in the caption following the image

Plots showing the x and y components of the geomagnetic perturbations produced by different current systems, the derivatives of the x and y components of the geomagnetic perturbations, and the simulated geomagnetically induced currents at the Hezhou substation.

As shown in Figure 8, the ionospheric Hall and Pedersen currents contribute little to the geomagnetic disturbance components (Bx and By) at low latitudes during the whole event, as expected. Therefore, for conciseness, the contributions from the ionospheric Hall and Pedersen currents to the derivatives of the x and y components of the geomagnetic perturbations at the Hezhou substation are not plotted in Figure 8. The FACs play the dominant role in producing By during most of the storm except at approximately 12:00 local time. The GM currents are predominantly responsible for generating Bx before 00:45 UT on 24 July; unexpectedly, however, after 00:45 UT on 24 July (between approximately 09:00 and 15:00 local time), the FACs play an important role in producing Bx. We consider this phenomenon to be dependent on the local time. In other words, the configuration of FACs and the local time at a particular geomagnetic station determine the contribution of FACs to the geomagnetic perturbations at that station.

Figure 9 shows the directionality of the geomagnetic disturbances generated by FACs from a global perspective in the form of a schematic diagram. In this figure, we plot only the Region 1 FACs because these FACs are dominant during the main phase of a geomagnetic storm, as shown in Figure 3. For conciseness, ring currents are not plotted. As shown in Figure 9, the geomagnetic disturbances generated by FACs are southward near 12:00 local time and northward close to 24:00. At dawn and dusk, the geomagnetic disturbances generated by FACs are oriented westward and eastward, respectively. Therefore, during the main phase of a geomagnetic storm, as the local time in the area of the power grid approaches 12:00, the southward geomagnetic disturbance produced by FACs is superimposed onto the southward geomagnetic disturbance produced by ring currents, thereby enhancing the southward geomagnetic disturbance there. This explains why we discovered that FACs begin to play an important role in producing Bx after 00:45 UT on 24 July between approximately 09:00 and 15:00 local time. The contribution of FACs to dBx/dt becomes very obvious after 00:45 UT on 24 July and is comparable to that of GM currents. For dBy/dt, the strength of dBy/dt is equivalent to that of dBx/dt, however, the contribution from FACs is obviously dominant. The large GIC peak at approximately 23:00 UT on 23 July corresponds to the large dBx/dt peak generated mainly by GM currents and to the dBy/dt peak generated mainly by FACs. In contrast, the large GIC peak at 00:58 UT on 24 July corresponds to the large dBx/dt peak generated by both GM currents and FACs and the dBy/dt peak generated mainly by FACs. These results shed light on the fact that in addition to ring currents, FACs play an important role in producing large GICs at low latitudes during geomagnetic storms.

Details are in the caption following the image

Schematic diagram of the geomagnetic disturbances generated by field-aligned currents from a global perspective during a geomagnetic storm. The coordinates on the right show the directions of the geocentric solar ecliptic coordinates.

The EEJs are important current systems that influence GICs in equatorial power grids (Carter et al., 2015). EEJs are located in a narrow daytime equatorial region (within ±3° of the magnetic equator; Yizengaw et al., 2014). The geomagnetic latitude of the lowest latitude substation (S50 in Figure 5) of Guangdong 500 kV power grid is about 9°. The EEJs are unlikely to influence the GICs in Guangdong 500 kV power grid. However, we should keep in mind that the simulation does not include the EEJs and solar quiet (Sq) current systems, which can influence GICs around the dip equator and low-latitude regions.

5 Discussion

In this study, the SWMF model is applied to calculate the geomagnetic perturbations at the Earth's surface. The simulation performance of the SWMF model has been thoroughly validated by many previous works (e.g., Pulkkinen et al., 2011; Yu et al., 2010; J. J. Zhang et al., 2020, and others). Recently, J. J. Zhang et al. (2020) simulated the GICs at a low-latitude substation during six storms based on the geomagnetic perturbations calculated from the SWMF model and showed that the SWMF can accurately reproduce the north component of geomagnetic perturbations (Bx). The highest cross-correlation (CC) coefficient between the observed and simulated Bx for the six events was 0.90, and the lowest normalized root-mean-square (nRMS) difference was 0.27. However, the ability of the model to reproduce By is weaker than that of Bx: the highest CC coefficient for By was only 0.46. Furthermore, the variations in Ex, Ey and GICs are highly complicated, so the CC coefficient and nRMS are not suitable for evaluating the results of the simulations. Nevertheless, according to the results, the simulations captured the main active periods of Ex, Ey and the GICs during each geomagnetic storm with intensities comparable to those observed. Furthermore, the event-based analysis showed that the SWMF model is more applicable than the persistence model in the prediction of GICs within low-latitude power grids during storms.

In this study, the 1-min sampling rate of the simulated B is used to model E. A higher sampling rate of B may increase the peak value of the modeled E and GICs. Grawe et al. (2018) quantified the effect of the sampling rate of a magnetometer on modeled peak surface electric field intensities during storms and demonstrated by using 1-D conductivity models that a 1-min sampling rate of magnetometer data leads to a 10%–20% magnitude attenuation of the peak electric field intensity across the United States.

Near the continental coastline, the electric field produced on the landward side is substantially enhanced due to the abrupt change in conductivity; this phenomenon is called the “coast effect” (Gilbert, 2005). Guangdong Province is a coastal region located in southeastern China. C. Liu et al. (2016) estimated the influence of the “coast effect” on the GICs flowing within the Guangdong 500 kV power grid and illustrated that the GICs at the Ling'ao substation (S43 in Figure 5 of this paper) would increase by 50% when taking the “coast effect” into account. However, in this study, a 1-D layered ground conductivity model of Guangdong Province is applied to model the geoelectric field without considering the “coast effect.” Therefore, the GICs calculated in this study for the ocean-adjacent substations in the Guangdong 500 kV power grid may be underestimated.

GICs are also affected by connections to neighboring systems. Four substations in the Guangdong 500 kV power grid are connected to the Guangxi power grid to the west: the Hezhou substation (S1 in Figure 5), Wuzhou substation (S2), Xianlingshan substation (S4), and Maoming substation (S27); these substations are treated as boundaries in the power grid simulation conducted in this study. Among these four substations, to cut off GIC flow paths, series compensation capacitors are installed on the Hezhou (S1)—Liudong Line, the Xianlingshan (S4)—Guilin Line, and the Maoming (S27)—Yulin Line. Therefore, the Hezhou, Xianlingshan and Maoming substations can be reasonably treated as boundaries in the power grid. In contrast, the Wuzhou substation (S2) connected to the Laibin substation in the Guangxi power grid is not equipped with a series compensation capacitor; thus, when this substation is treated as a boundary, the GICs at this substation can be overestimated. Boteler et al. (2013) provided an equivalent circuit method for modeling GICs from a neighboring network. Accordingly, adding equivalent circuits at the Wuzhou (S2) substation would improve the accuracy of GIC calculations at this substation in future work.

6 Summary

The risk of a power grid to GICs during solar superstorms is an issue of great concern. Specifically, research has not quantified how large such GICs could become in a real power grid affected by a solar superstorm. In this study, we combined the SWMF model with a ground conductivity model and a model of the Guangdong 500 kV power grid to evaluate the response of the whole power grid if the 23 July 2012 solar superstorm had struck the Earth. We then assessed the GIC risk of the power grid. According to the results, the maximum |GIC| reached approximately 400 A in the Guangdong 500 kV power grid during the solar superstorm, and this magnitude is more than three times that of the GICs produced during a strong magnetic storm with Kp = 8. A statistical analysis of the GICs flowing through the substations in this power grid showed that the maximum GICs exceeded 50 A at 74% of the substations, and the total duration of GICs whose magnitudes exceeded 30 A was longer than 30 min at 55% of the substations during the storm. On the basis of the calculated strengths of the GICs, the durations of large GICs and a comparison with the GICs responsible for the Hydro-Québec power outage that occurred on 13 March 1989, we assert that the risk of GICs in the Guangdong 500 kV power grid will be very high in the event that a solar superstorm strikes the Earth. This indicates that even a low-latitude power grid may collapse under the influence of a solar superstorm. Thus, power grids worldwide should be protected against the GIC risks caused by solar superstorms, and relevant officials should implement disaster prevention and reduction plans to prepare for such an emergency. On the other hand, we analyzed the contributions of different current systems to GICs at low latitudes and found that FACs can play an important role in producing large GICs in such areas during geomagnetic storms. This finding provides crucial insight for understanding the factors that produce strong GICs at middle and low latitudes.

Acknowledgments

The authors acknowledge the use of solar wind data from STEREO-A, which can be obtained from CDAWeb (https://cdaweb.gsfc.nasa.gov/index.html/). This work was supported by National Natural Science Foundation of China grants 41731070, 41774155, 42174210, 41774179, and 41821003 and by the Strategic Pioneer Program on Space Science, Chinese Academy of Sciences (Grant No. XDA15052500). This work was also supported in part by the Specialized Research Fund for the State Key Laboratories of China and by the Chinese Meridian Project. The simulations were performed on TianHe-2 at the National Supercomputer Center in Guangzhou, China.

    Data Availability Statement

    The SWMF input files are openly accessible online (https://doi.org/10.5281/zenodo.6072061). The GIC calculation code can also be downloaded online (https://doi.org/10.5281/zenodo.6099318).