Simulation of Geomagnetically Induced Currents: Past, Present and Future Events Magnus Wik 1, Risto Pirjola 2, Ari Viljanen 2, Henrik Lundstedt 1, Peter Wintoft 1, Antti Pulkkinen 3 1 Swedish Institute of Space Physics, Lund, Sweden 2 Finnish Meteorological Institute, Helsinki, Finland 3 NASA Goddard Space Flight Center, Greenbelt, Maryland, USA Third European Space Weather Week Brussels, Belgium November 13-17, 2006

Abstract Geomagnetically induced currents (GIC) flowing in technological networks at the Earth’s surface, such as electric power transmission systems, oil and gas-pipelines, telecommunication cables and railway equipment, are the ground end of the complicated space weather chain that originates from solar activity. GIC are generally a source of problems to the systems. The world record GIC (~300 A) was measured in the earthing lead of a transformer neutral in southern Sweden during the magnetic storm on April 6, On October 30, 2003, the city of Malmö at the southern coast of Sweden suffered from a power blackout due to GIC. This outage affected customers and lasted between 20 to 50 minutes. We present some results of calculations which identify the sites in the 400 kV power system in southern Sweden that most probably experience large GIC values. We show comparisons of calculated and measured GIC during magnetic storm events from the present solar cycle. A calculation of GIC on today’s power grid for the Carrington- and Halloween event is also shown. We also discuss what we can expect from future events. Data The data necessary for the simulation of GIC consists of power grid data and magnetic field data. The power grid data is divided into: transformer station coordinates, line resistances, transformer resistances and earthing resistances. We only use data for the 400 kV part. Most of the data comes from SVK (Svenska kraftnät) but for a few stations we used averaged values for resistances and coordinates. These stations where however located far from the station where the measurements took place (SVP-2 in the map.) The network is shown in Fig. 1. The magnetic field was recorded at several sites in northern Europe (Fig. 2). The magnetic field was then interpolated on a grid covering southern Sweden. Using a 1-D conductivity model the electric field was then calculated using the local plane wave method. The grid covering southern Sweden is shown in Fig. 3. The measured GIC (1 min resolution) data was collected during parts of the period at Simpevarp-2.

Fig. 1: The 400 kV network in southern Sweden. The grid consists of 22 substations and 24 transmission lines. The station coordinates are only approximate. Station Sege is close to Malmö. Measured GIC was collected at Simpevarp 2 (21 on the map). Fig. 2: Magnetometer stations.Fig. 3: The dense grid covering southern Sweden.

Calculation of GIC GIC modelling is divided into two independent parts: 1. Determination of the geoelectric field. 2. Calculation of GIC using the given geoelectric field. Induced electric field GIC/phase R transf. R transmission line Neutral point Electrojet Fig. 4: GIC flowing between 2 substations. Geomagnetic recordings from stations nearby southern Sweden (Fig. 2) was used. They included all IMAGE magnetometer array sites and seven INTERMAGNET stations. The geomagnetic data was interpolated to a spatially denser grid covering the area for the 400 kV system where the GIC modeling was carried out (Fig. 3). Using the Spherical Elementary Current System (SECS) method an ionospheric equivalent current system was derived for each time step. From the equivalent current system, the magnetic field was then computed at several points on the surface of the Earth. The computed field satisfies the physical properties of the geomagnetic field. The geoelectric field is then computed by the local plane wave method on the same grid. The electric field depends on all previous values of the magnetic field, although the most recent have the largest effect. The induced electric field causes current to flow in the ground and up through the transformer grounding to the neutral point where it follow the transmission lines down to the next substation. These ~DC currents can cause damage to the transformer (Fig. 4)

Results Fig. 5: Measured and calculated GIC for two events from the present solar cycle.

Results Fig. 6: Calculated GIC for part of the Halloween event and Carrington event.

We calculated GIC in the Swedish 400 kV system due to an electric field of 1V/km having any direction. In Fig. 7 is a plot showing the geoelectric field direction that gives the largest GIC at each station. It is clear from the figure that the largest GIC is found at the corners of the network. During geomagnetic storms we can therefore expect substations located at the corners and ends of the power grid to experience larger GIC. Measured GIC was compared to calculated GIC for two events (Fig. 5). The calculated GIC follow the measured during the whole event except during parts with high peaks in the current. As a test we trained a simple linear neural network with only the electric field as an input and measured GIC as target. The output was similar to the calculated and also had the same correlation with the measured GIC. This might indicate that a 2-D or 3-D conductivity model is needed. However it is possible that improved power grid data is needed at least close to Simpevarp, where GIC was measured. Conclusion Next we calculated GIC for the Halloween event, October 29, The agreement to measured GIC seems to be rather good. Finally we calculated GIC based on magnetometer data recorded at Kew observatory on September 2, 1859 (The Carrington event). This result should be regarded as very approximate. Since this part of the storm was not the most disturbed it is likely that much higher values for GIC is expected in today’s power system if a new “Carrington event” occur. Fig. 7: This plot is showing the geoelectric field direction which causes largest current for each station. The length of the arrows is the maximum GIC for that station for an electric field of 1 V/km

Acknowledgements We wish to thank Prof. Sture Lindahl (Gothia Power AB, Lund, Sweden) for useful discussions and advice about power grid modelling in GIC calculations. We are grateful to ELFORSK, E.ON and Svenska Kraftnät (SVK) for supporting our studies and for providing power grid data to us. This study is partly included in the ESA pilot project SDA: “Real-time forecast service for geomagnetically induced currents” supported by ESA and ELFORSK. That study was a joint collaboration between Swedish Institute of Space Physics (IRF) and Finnish Meteorological Institute (FMI). Our thanks also goes to Dr. David Boteler (NRCan, Ottawa, Canada) for discussions about past geomagnetic storms and for providing digitised magnetic data of the Carrington storm. In the calculation of GIC we partly neglected the lower voltage, 135 kV, that in reality is connected to the 400 kV system. We also used a 1-D conductivity model for calculating the electric field. In the near future we plan to include the 135 kV system as well as a 2-D conductivity model. We will also compare measured and calculated GIC with different wavelet methods to reveal how coherent the two time series are in frequency and power. Even though there are no general agreement about solar cycle 24 we can expect many different kind of GIC events. During past events we have seen geomagnetic storms produced by interacting CMEs (e.g. during Halloween and in 1989), events from CMEs produced within a coronal hole and coronal holes. During the next Gleissberg maximum, in ~year 2030, stronger geomagnetic storms are expected and larger GICs. Since it is very difficult for power companies to mitigate effects from solar activity - geomagnetic storms it is necessary to make simulations and forecasts of GIC. The spacecrafts STEREO was just released and will hopefully help us provide better forecasts of GIC and space weather events.