An Overview of the U.S. Power Grid Model for the Geomagnetic Storm Threat EnvironmentsThe ability to comprehensively assess the vulnerability of the U.S. power grid to the geomagnetic storm environment produced by solar activity stems from the parallel investigations that have been underway to understand the problems of power system vulnerability from high altitude nuclear-burst (HEMP) events. Power system impacts from geomagnetic storms were first observed in 1940 and have been growing in importance as the power system has grown over the intervening years. Geomagnetic storms are created when the Earth's magnetic field captures ionized particles carried by the solar wind due to coronal mass ejections or coronal holes at the Sun. Although there are different types of disturbances noted at the Earth surface, the disturbances can be characterized as a very slowly varying magnetic field, with rise times as fast as a few seconds, and pulse widths of up to an hour. The rate of change of the magnetic field is a major factor in creating electric fields in the Earth and thereby inducing quasi-dc current flow in the power transmission network. Unlike the HEMP threats, geomagnetic storms are a much more frequent occurrence, which also allows for extensive opportunities to fully benchmark each component of the simulation models and therefore provide greater confidence in the analysis of plausible severe threats, such as the threat posed by an extreme geomagnetic storm scenario. The context of the evolution of power system discoveries of vulnerability can be further understood by an overview of the geomagnetic storm phenomena, which is closely associated with the more familiar variability of the sunspot cycle. Figure 1-1 provides a plot of the sunspot count as well as large geomagnetic storms over the last 70 years. Because each sunspot cycle is typically ~11 years in duration, this plot provides the status of the current solar cycle (Cycle 23) back to Cycle 17, which began in ~1932. As can be seen, not all sunspot cycles are of equal intensity and Cycle 19 (late 1950’s-early 1960’s) is in fact the largest sunspot cycle of human-record. The most recent cycle, Cycle 23, exhibits a profile similar to that of Cycle 17. As noted in this figure, at the time of Cycle 19, much of the present U.S. power grid high-voltage transmission system of today did not exist. To further explore the importance of the evolution of the power grid and its growing vulnerability, it is necessary to look at specific large geomagnetic storms, as the sunspot count does not provide sufficient correlations to impacts observed at the Earth. In classification of the intensity of geomagnetic storms an index called the Ap index is used, which provides a planetary measure of storm activity. Many of these storms caused notable impacts to various terrestrial technology systems of their respective eras. The storms of March 1989 and several in 1991 produced large and unprecedented operational impacts to power grids in the U.S. and at other world locations. Also noteworthy is that large geomagnetic storms generally have not occurred around the peaks of sunspot activity. For example, a storm in February 1986 caused power system problems all across the eastern U.S. but actually occurred at the absolute minimum between Solar Cycle 21 and 22. Because sunspots only provide a gross measure of overall solar activity, it does not accurately reflect the discrete eruptive events from violent solar active regions that, when Earth-directed, can trigger large geomagnetic storms. Rather, it is clear from this comparison that large and threatening geomagnetic storms can occur at any time during the sunspot cycle, and pose a near continuous threat probability. When reviewing the occurrence of large storms, it is important to recognize that the problem of power system impacts is compounded by growing vulnerability of this infrastructure to geomagnetic disturbances. The extent of the growth in vulnerability over time is due to factors stemming from the growth of the high-voltage transmission grid in the U.S., as well as changes within the grid that introduce new or enhance existing impact problems to the power grid. Figure 1-2 shows the growth of the U.S. high voltage transmission grid over the last 50 years. This geographically widespread infrastructure readily couples through multiple ground points to the geo-electric field produced by disturbances in the geomagnetic field. As shown, from Cycle 19 through Cycle 22, the high voltage grid grew nearly tenfold. In essence, the antenna that is sensitive to disturbances has grown dramatically over time. As this network has grown in size, it has also grown in complexity. As will be discussed in later sections, one of the more important changes in the technology base for the U.S. power grid that can increase impacts to geomagnetic storms is the evolution to higher operating voltages of the network. The operating levels of the high voltage network has increased from the 115- 230kV levels of the 1950’s to networks that operate from 345kV, 500kV and 765 kV across the continent. In order to quantify the impacts of the severe geomagnetic storm threats to the U.S. power grid is it necessary to develop a series of models that translate the disturbed space environment, or geomagnetic field environment, into specific impacts to the operation of the electric power grid. This requires the following steps: • Modeling in detail the geographically wide-spread disturbances to the geomagnetic field from natural geomagnetic storm processes. • Modeling the electromagnetic coupling between the disturbed space environment and the deep-earth ground conductivity that produces a geo-electric field across the surface of the Earth. • Modeling the interaction between the geo-electric field and the complex power grid topology to calculate the flow of geomagnetically induced current (GIC) throughout the exposed power grid infrastructure. • Modeling of the operational impacts in the U.S. power grid due to GIC flows caused by either E3 threats or severe geomagnetic storm conditions. While each of these models and associated environments are complex, these modeling efforts have been highly successful in accurately replicating geomagnetic storm events and performing detailed forensic analysis of geomagnetic storm impacts to electric power systems. This capability has also been successfully applied towards providing predictive geomagnetic storm forecasting services to the electric power industry. To further describe the methodology used in this analysis, a brief overview is provided for each of the key modeling steps that were undertaken. An important facet of this investigation requires the simulation of geomagnetic storm events and the impacts that these storms caused to electric power grid operation, and to also investigate the potential impacts of very large storm events that have not recently been experienced by today’s power systems. Electric power system operators realize that large and severe geomagnetic storms (such as the March 1989 storm) have the potential to cause important power system impacts. However, the U.S. power industry in general have not developed comprehensive simulation models such as being developed in this effort to better quantify the nature of the threat environment. The power industry also has a very limited perspective on the extremes of storm intensity due to the flaws of the K Index rating of storms. While some past storms have severely threatened the U.S. grid, these storms do not represent the most severe storm events that are plausible. Therefore, comprehensive models allow the development of improved understandings of the extremes of the geomagnetic environment and consequential impacts that future severe storms may pose to the integrity of this important infrastructure... Geomagnetic disturbances are caused by interactions of the solar wind with the Earth’s magnetic field. There are a number of ways that a geomagnetic storm can produce a ground-level geomagnetic field disturbance that could have the potential to impact power system operations. One of the most important geomagnetic storm processes involves the intensification and flow of ionospheric currents known as electrojets. These electrojets are formed around the north and south magnetic poles at altitudes of about 100km and can have magnitudes of ~1 million amps, which is sufficient in intensity to cause widespread disturbances to the geomagnetic field. Because of the large geographic scale of the U.S. power grid, it would not be suitable to assume the application of a simple planewave model for the disturbance conditions. Therefore, to simulate the geomagnetic storm environment, it is necessary to develop a geographically gridded specification of the complex spatial and temporal dynamics of the disturbances as they propagate across North American locations that are to be modeled.... Figure 1-3 shows the vector description of a disturbance of the geomagnetic field simultaneously observed at a number of locations across North America at time 9:10 UT May 10, 1992. ...Ground conductivity models need to accurately reproduce geo-electric field variations that are caused by the very low frequency ranges of geomagnetic storms. These electromagnetic disturbances require models accurate over a frequency range from 0.3 Hz to as low as 0.00001 Hz. Because of the low-frequency content of the disturbance environments, it is necessary to take into consideration ground conductivities to appropriate depths... These conductivity variations with depth can range 3 to 5 orders of magnitude. While surface conductivity can exhibit considerable lateral heterogeneity across the U.S., conductivity at depth is more uniform. Because of this, models of ground conductivity can be successfully applied over meso-scale distances and can be accurately represented by use of layered conductivity profiles or models. Frequent occurrences of geomagnetic storm events and subsequent measurement of these storm environments and associated impacts have provided opportunities to use this information to develop and validate models of the ground conductivity for the U.S. Power Grid Model. A severe electrojet disturbance can produce a rate-of-change intensity profile (or dB/dt) of 2400 nT/min or greater. This disturbance is a very severe disturbance that could be possible at high to mid latitude locations throughout the U.S... a map of the overall transmission network included in the CONUS region model of the U.S. grid. The major transmission voltages are color highlighted by operating voltage with the three operating voltages of 345kV, 500kV and 765kV (there are also several lines in the Washington state region that are operated at 300kV which are included in the model, and in the figure are combined with 345kV lines). Figure 1-10 provides the mileage statistics for each of these three voltage classes that are represented in the U.S. model for the CONUS region. As shown, the most common transmission voltage is the 345kV, which makes up about 64% of total transmission line miles. The highest operating voltage is the 765kV and is primarily located in the Illinois, Ohio, Indiana, West Virginia and upstate New York regions of the U.S. Both the 345kV and 500kV portions of the network are more widely distributed across the U.S... Figure 1-9. Map of 345kV, 500kV and 765kV substations and transmission network in U.S. grid model. The operating voltage of the transmission network is an important factor in determining the level of GIC flow that will occur on each part of the U.S. power grid. At the higher operating voltages, there are pronounced trends that: the average length of each line increases and the average circuit resistance decreases. These trends result in larger GIC flows in the higher voltage portions of the network, given the same geo-electric field conditions... Transformers also exhibit a general characteristic of lower resistance as kV rating increases. However, the trend in transformer design is even more pronounced as a function of MVA or current rating of the transformer. It is generally consistent that the higher MVA rated transformers are also the highest kV rated transformers, though there can be a few exceptions. In addition to the lower resistances at the higher kV rating lines on the network, average length of these lines also introduces a higher overall risk of GIC flows as well. Figure 1- 14 provides a summary of average transmission line lengths in the U.S. by kV rating. As illustrated, the average length of transmission lines also increases significantly with increased kV ratings. The 765kV lines average over 60 miles in length while the 115kV lines are less than 15 miles in average length. While predicting GIC flows, it is necessary to take into consideration the network topology as a integrated whole. It is evident that on an individual line basis a combination of longer average length (and increased geoelectric potential between end points of the line) combined with lower average resistances will produce substantially larger GICs on average in the higher voltage portions of the power grid. |
The flow of GIC in transformers is the root cause of all
power system problems, as the
GIC causes half-cycle saturation to occur in the exposed transformers. While the
extremes of the threat environment, the conductivity of the deep-earth ground,
and the kV
rating and topology of the power grid can all cause significant enhancements of
the total
GIC flows, the most significant enhancement of impacts due to GIC is how that
GIC
interacts within the transformer. Only a few amps of GIC can result an
amplification of
impacts in the operation of AC current flows in the transformer. In some cases
the
amplification effect can cause normal AC excitation current in a transformer to
increase
from less than 1 amp to nearly 300 amps, due to the flow of only 25 amps/phase
of GIC....
The discussion to this point has been primarily
regarding the threat to reliability of the
system as a whole, due to the widespread and simultaneous nature of the stress
across a
large interconnected network caused by the March 1989 storm. Also of note from
this
particular storm is strong evidence that GIC-induced half-cycle saturation of
transformers
can indeed produce enough heat to severely damage or even destroy exposed large
power
transformers...
In looking at the solar source, the size of flares (as
measured by X-ray emissions)
provides one of the best and longest recorded classification methods. Figure 3-7
provides
a plot of the observed large flares since 1972 (when reliable X-Ray observations
of solar
flares became available). The size scale of flares used by the NOAA Space
Environment
Center is a classification of M and X based on logarithmic decade change in the
X-ray
energy observed. There are several flares that have exceeded even the X category
by
another factor of 10, and are classified as X+. This is also a range in which
instrument
saturation begins to limit accuracy in determining total energy content of
flares.
However, the flare that was suspected of triggering the March 13, 1989
superstorm was
only in the mid range of the X class, and not close to the most energetic events
observed.
What is more relevant is the location of the eruption on the solar disk and the
resulting
CMEs probable connection to the Earth. In particular, the large X22+ flare event
of April
2, 2001, while ~30 times larger than the March 89 flare, was located at the far
west limb
of the Sun, and the resulting CME ejecta was not Earth-directed and only
provided a
small glancing blow upon arrival at the Earth.
As the ejecta leaves the Sun, a structure that is
commonly called a coronal mass ejection
(CME) begins to travel through interplanetary space, and generally in the
direction of
initial acceleration from the source. The CME, by the time it arrives at Earth,
can be a
massive structure, with a cross-sectional dimension measured along the Sun-Earth
line
that can be on the order of 0.5 AU or greater. These CMEs contain a magnetic
structure
which greatly enhances the Interplanetary Magnetic Field (IMF). The IMF
intensity and
orientation are important in defining the nature of the interaction with the
Earth’s
magnetosphere. The most favorable orientation to produce a storm is when the IMF
Bz
component is negative, or southward oriented, and therefore opposite of the
Earth’s
bipolar field. Under these conditions, a reconnection readily occurs between the
IMF and
the Earth’s magnetosphere, allowing particles to enter and greatly enhance
geomagnetic
storm processes. IMF speed is also an important contributor, as well, to
defining the
level of energetics that the solar wind IMF delivers to the magnetosphere. The
ability to
continuously measure the solar wind has only been established over the past few
years;
therefore observations of this type are not available for the March 13-14, 1989
superstorm. However, one of the larger storms in the just completed solar cycle
was
observed on July 15-16, 2000. The nature of the coupling between the solar wind
and the
Earth’s magnetosphere can be further illustrated. Figure 3-8 provides a plot of
the
rectified electric field (a measure of storm energetics) for a large CME cloud
passage
during July 15-16, 2000. Two plot areas are shown. In blue is a plot of the
total solar
wind energy, which is based on speed and B total of the IMF. As previously
explained, it
is only when Bz of the IMF is pointing southward that coupling with the
magnetosphereoccurs. The red plot provides this fraction of the total solar wind
cloud content that
couples and produces storm activity. Because the solar wind IMF exhibited a
bi-polar
rotation of Bz during the passage of the cloud, only a fraction of total energy
was able to
couple. Figure 3-9 again shows the coupled solar wind energy, overlaid by a plot
of the
observed ground level geomagnetic field disturbances observed at Fredericksburg,
MD
(in nT/min) during the storm. This comparison indicates a fairly close coupling
of the
intensity variations over the storm interval. The July 15-16, 2000 event had
many of the
solar wind features that are considered to be approaching upper bounds. However,
the
storm, in total, was limited by the coupling efficiency. One way of measuring
this
coupling efficiency is to accumulate the energy content of the total solar wind
and the
coupled portion. In actuality, the coupling efficiency for this cloud passage
was only at
~40% of total solar wind content. Figure 3-10 shows the solar wind energy
content
comparisons (total and coupled) for this storm as well as other noteworthy
storms of 2000
and 2001. This summary indicates that the July 2000 storm had the highest solar
wind
content of the storms examined, and also the highest coupled content, with
considerable
size variations possible. It also indicates that the percentage coupling
efficiency can be
quite variable. For example, in the events considered here, the coupling
efficiency ranges
from a low of ~3% to a high of ~85%.
The highest intensity portion of the cloud has a period
of ~12-18 hours, as this cloud
event was due to a single large CME ejecta from the Sun. There are important
contrasts
when looking at the duration of great storms, such as March 13-14, 1989, which
lasted
over 24 hours. The March 89 storm also had the most energetic substorms in late
March
13 and into March 14 after about a 12 hour lull from substorms in the earlier AM
hours of
March 13. This suggests the possibility of passage of a second CME cloud that
triggered
the late energetic substorms. The magnetosphere has several modes of inertia and
storage
that play an important role in very long duration storms.
...
These at-risk transformers also represent a diverse
population of function and high and
low side kV and MVA ratings. The at-risk populations are made up of auto and
non-auto
transformer types with a variety of primary and secondary voltage ratings and
MVA
capacity ratings that were designed specific to their grid location purposes.
This diversity
underscores the problems of providing spare equipment for such large scale
infrastructure
failures. Also from a world market manufacturing perspective, these numbers of
failures
exceeds the annual production in the world of transformers of this kV rating and
MVA
size class. Normally, only a handful of transformers of this size are purchased
for U.S.
locations on an annual basis. Therefore the immediate replacement of such a
large scale
failure of the infrastructure could pose serious challenges and add considerable
delays to
the restoration process for the power grid.
Of particular concern would be the permanent loss of large GSU (generator
step-up)
transformers at power plants in the northeastern region of the U.S. (i.e. NE
Quad). The
loss of these transformers causes a compounding of difficulties, in that the EHV
transmission network is impaired along with the loss of output of vital and
usually
baseload nuclear, coal, and hydro-electric generation resources for the power
grid. There
are a considerable number of the large GSU transformers “at-risk” due to GICs of
at least
30 amps per phase in these units.
Overview of Emergency Replacement of EHV Transformers
The failure of many large EHV transformers and the need to suddenly replace a
large
number of them has not been previously contemplated by the U.S. electric power
industry. Under normal conditions, the purchase placement of a single EHV
transformer
order in the 300-400MVA class has normally been quoted as taking up to 15 months
for
manufacture and test. For larger sizes of transformers and transformers with
special
reactance or tap-changer requirements, several months may need to be added to
the above
mentioned figure, and the suitability of qualified manufacturers may be more
limited.
Of course, manufacturing and testing the equipment does not mean the story ends
there.
The equipment will then need to be transported to site and commissioned before
being
put into service. The size and weight of large EHV transformers precludes the
concept of
airlifting from an overseas destination for emergency replacements, even if a
suitable
spare is readily available. This means at least several weeks of ocean transport
for
apparatus of foreign source. When such heavy equipment arrives at the border or
port, italmost always requires permission from municipalities and
highway/transport authorities,
as they are slow moving and heavy...
http://www.fas.org/irp/eprint/geomag.pdf